# Applied Machine Learning Online Course

#### How to utilise Appliedaicourse

#### Python for Data Science Introduction

- Python, Anaconda and relevant packages installations
- Why learn Python?
- Keywords and identifiers
- comments, indentation and statements
- Variables and data types in Python
- Standard Input and Output
- Operators
- Control flow: if else
- Control flow: while loop
- Control flow: for loop
- Control flow: break and continue

#### Python for Data Science: Data Structures

#### Python for Data Science: Functions

#### Python for Data Science: Numpy

#### Python for Data Science: Matplotlib

#### Python for Data Science: Pandas

#### Python for Data Science: Computational Complexity

#### Plotting for exploratory data analysis (EDA)

exploratory data analysis (EDA) is an approach to analyzing data sets to summarize their main characteristics, often with visual methods. A statistical model can be used or not, but primarily EDA is for seeing what the data can tell us beyond the formal modeling or hypothesis testing task.

- Introduction to IRIS dataset and 2D scatter plot
- 3D scatter plot
- Pair plots
- Limitations of Pair Plots
- Histogram and Introduction to PDF(Probability Density Function)
- Univariate Analysis using PDF
- CDF(Cumulative Distribution Function)
- Mean, Variance and Standard Deviation
- Median
- Percentiles and Quantiles
- IQR(Inter Quartile Range) and MAD(Median Absolute Deviation)
- Box-plot with Whiskers
- Violin Plots
- Summarizing Plots, Univariate, Bivariate and Multivariate analysis
- Multivariate Probability Density, Contour Plot
- Exercise: Perform EDA on Haberman dataset

#### Linear Algebra

It will give you the tools to help you with the other areas of mathematics required to understand and build better intuitions for machine learning algorithms.

- Why learn it ?
- Introduction to Vectors(2-D, 3-D, n-D) , Row Vector and Column Vector
- Dot Product and Angle between 2 Vectors
- Projection and Unit Vector
- Equation of a line (2-D), Plane(3-D) and Hyperplane (n-D), Plane Passing through origin, Normal to a Plane
- Distance of a point from a Plane/Hyperplane, Half-Spaces
- Equation of a Circle (2-D), Sphere (3-D) and Hypersphere (n-D)
- Equation of an Ellipse (2-D), Ellipsoid (3-D) and Hyperellipsoid (n-D)
- Square ,Rectangle
- Hyper Cube,Hyper Cuboid
- Revision Questions

#### Probability and Statistics

- Introduction to Probability and Statistics
- Population and Sample
- Gaussian/Normal Distribution and its PDF(Probability Density Function)
- CDF(Cumulative Distribution function) of Gaussian/Normal distribution
- Symmetric distribution, Skewness and Kurtosis
- Standard normal variate (Z) and standardization
- Kernel density estimation
- Sampling distribution & Central Limit theorem
- Q-Q plot:How to test if a random variable is normally distributed or not?
- How distributions are used?
- Chebyshev’s inequality
- Discrete and Continuous Uniform distributions
- How to randomly sample data points (Uniform Distribution)
- Bernoulli and Binomial Distribution
- Log Normal Distribution
- Power law distribution
- Box cox transform
- Applications of non-gaussian distributions?
- Co-variance
- Pearson Correlation Coefficient
- Spearman Rank Correlation Coefficient
- Correlation vs Causation
- How to use correlations?
- Confidence interval (C.I) Introduction
- Computing confidence interval given the underlying distribution
- C.I for mean of a normal random variable
- Confidence interval using bootstrapping
- Hypothesis testing methodology, Null-hypothesis, p-value
- Hypothesis Testing Intution with coin toss example
- Resampling and permutation test
- K-S Test for similarity of two distributions
- Code Snippet K-S Test
- Hypothesis testing: another example
- Resampling and Permutation test: another example
- How to use hypothesis testing?
- Proportional Sampling
- Revision Questions

#### Interview Questions on Probability and statistics

#### Dimensionality reduction and Visualization:

In machine learning and statistics, dimensionality reduction or dimension reduction is the process of reducing the number of random variables under consideration, via obtaining a set of principal variables. It can be divided into feature selection and feature extraction.

- What is Dimensionality reduction?
- Row Vector and Column Vector
- How to represent a data set?
- How to represent a dataset as a Matrix.
- Data Preprocessing: Feature Normalisation
- Mean of a data matrix
- Data Preprocessing: Column Standardization
- Co-variance of a Data Matrix
- MNIST dataset (784 dimensional)
- Code to Load MNIST Data Set

#### PCA(principal component analysis)

- Why learn PCA?
- Geometric intuition of PCA
- Mathematical objective function of PCA
- Alternative formulation of PCA: Distance minimization
- Eigen values and Eigen vectors (PCA): Dimensionality reduction
- PCA for Dimensionality Reduction and Visualization
- Visualize MNIST dataset
- Limitations of PCA
- PCA Code example
- PCA for dimensionality reduction (not-visualization)

#### (t-SNE)T-distributed Stochastic Neighbourhood Embedding

#### Interview Questions on Dimensionality Reduction

#### Real world problem: Predict rating given product reviews on Amazon

- Dataset overview: Amazon Fine Food reviews(EDA)
- Data Cleaning: Deduplication
- Why convert text to a vector?
- Bag of Words (BoW)
- Text Preprocessing: Stemming, Stop-word removal, Tokenization, Lemmatization.
- uni-gram, bi-gram, n-grams.
- tf-idf (term frequency- inverse document frequency)
- Why use log in IDF?
- Word2Vec.
- Avg-Word2Vec, tf-idf weighted Word2Vec
- Bag of Words( Code Sample)
- Text Preprocessing( Code Sample)
- Bi-Grams and n-grams (Code Sample)
- TF-IDF (Code Sample)
- Word2Vec (Code Sample)
- Avg-Word2Vec and TFIDF-Word2Vec (Code Sample)
- Assignment-2: Apply t-SNE

#### Classification And Regression Models: K-Nearest Neighbors

- How “Classification” works?
- Data matrix notation
- Classification vs Regression (examples)
- K-Nearest Neighbours Geometric intuition with a toy example
- Failure cases of KNN
- Distance measures: Euclidean(L2) , Manhattan(L1), Minkowski, Hamming
- Cosine Distance & Cosine Similarity
- How to measure the effectiveness of k-NN?
- Test/Evaluation time and space complexity
- KNN Limitations
- Decision surface for K-NN as K changes
- Overfitting and Underfitting
- Need for Cross validation
- K-fold cross validation
- Visualizing train, validation and test datasets
- How to determine overfitting and underfitting?
- Time based splitting
- k-NN for regression
- Weighted k-NN
- Voronoi diagram
- Binary search tree
- How to build a kd-tree
- Find nearest neighbours using kd-tree
- Limitations of Kd tree
- Extensions
- Hashing vs LSH
- LSH for cosine similarity
- LSH for euclidean distance
- Probabilistic class label
- Code Sample:Decision boundary .
- Code Sample:Cross Validation
- Revision Questions

#### Interview Questions on K-NN(K Nearest Neighbour)

#### Classification algorithms in various situations

- Introduction
- Imbalanced vs balanced dataset
- Multi-class classification
- k-NN, given a distance or similarity matrix
- Train and test set differences
- Impact of outliers
- Local outlier Factor (Simple solution :Mean distance to Knn)
- k distance
- Reachability-Distance(A,B)
- Local reachability-density(A)
- Local outlier Factor(A)
- Impact of Scale & Column standardization
- Interpretability
- Feature Importance and Forward Feature selection
- Handling categorical and numerical features
- Handling missing values by imputation
- curse of dimensionality
- Bias-Variance tradeoff
- Intuitive understanding of bias-variance.
- Revision Questions
- best and wrost case of algorithm

#### Performance measurement of models

#### Interview Questions on Performance Measurement Models

#### Naive Bayes

- Conditional probability
- Independent vs Mutually exclusive events
- Bayes Theorem with examples
- Exercise problems on Bayes Theorem
- Naive Bayes algorithm
- Toy example: Train and test stages
- Naive Bayes on Text data
- Laplace/Additive Smoothing
- Log-probabilities for numerical stability
- Bias and Variance tradeoff
- Feature importance and interpretability
- Imbalanced data
- Outliers
- Missing values
- Handling Numerical features (Gaussian NB)
- Multiclass classification
- Similarity or Distance matrix
- Large dimensionality
- Best and worst cases
- Code example
- Assignment-4: Apply Naive Bayes
- Revision Questions

#### Logistic Regression

- Geometric intuition of Logistic Regression
- Sigmoid function: Squashing
- Mathematical formulation of Objective function
- Weight vector
- L2 Regularization: Overfitting and Underfitting
- L1 regularization and sparsity
- Probabilistic Interpretation: Gaussian Naive Bayes
- Loss minimization interpretation
- hyperparameters and random search
- Column Standardization
- Feature importance and Model interpretability
- Collinearity of features
- Test/Run time space and time complexity
- Real world cases
- Non-linearly separable data & feature engineering
- Code sample: Logistic regression, GridSearchCV, RandomSearchCV
- Assignment-5: Apply Logistic Regression
- Extensions to Generalized linear models

#### Linear Regression

#### Solving Optimization Problems

- Differentiation
- Online differentiation tools
- Maxima and Minima
- Vector calculus: Grad
- Gradient descent: geometric intuition
- Learning rate
- Gradient descent for linear regression
- SGD algorithm
- Constrained Optimization & PCA
- Logistic regression formulation revisited
- Why L1 regularization creates sparsity?
- Assignment 6: Implement SGD for linear regression
- Revision questions

#### Interview Questions on Logistic Regression and Linear Regression

#### Support Vector Machines (SVM)

- Geometric Intution
- Mathematical derivation
- Why we take values +1 and and -1 for Support vector planes
- Loss function (Hinge Loss) based interpretation
- Dual form of SVM formulation
- kernel trick
- Polynomial Kernel
- RBF-Kernel
- Domain specific Kernels
- Train and run time complexities
- nu-SVM: control errors and support vectors
- SVM Regression
- Cases
- Code Sample
- Assignment-7: Apply SVM
- Revision Questions

#### Interview Questions on Support Vector Machine

#### Decision Trees

- Geometric Intuition of decision tree: Axis parallel hyperplanes
- Sample Decision tree
- Building a decision Tree:Entropy
- Building a decision Tree:Information Gain
- Building a decision Tree: Gini Impurity
- Building a decision Tree: Constructing a DT
- Building a decision Tree: Splitting numerical features
- Feature standardization
- Building a decision Tree:Categorical features with many possible values
- Overfitting and Underfitting
- Train and Run time complexity
- Regression using Decision Trees
- Cases
- Code Samples
- Assignment-8: Apply Decision Trees
- Revision Questions

#### Interview Questions on decision Trees

#### Ensemble Models

- What are ensembles?
- Bootstrapped Aggregation (Bagging) Intuition
- Random Forest and their construction
- Bias-Variance tradeoff
- Train and run time complexity
- Bagging:Code Sample
- Extremely randomized trees
- Random Forest :Cases
- Boosting Intuition
- Residuals, Loss functions and gradients
- Gradient Boosting
- Regularization by Shrinkage
- Train and Run time complexity
- XGBoost: Boosting + Randomization
- AdaBoost: geometric intuition
- Stacking models
- Cascading classifiers
- Kaggle competitions vs Real world
- Assignment-9: Apply Random Forests & GBDT
- Revision Questions

#### Featurization and Feature engineering.

- Introduction
- Moving window for Time Series Data
- Fourier decomposition
- Deep learning features: LSTM
- Image histogram
- Keypoints: SIFT.
- Deep learning features: CNN
- Relational data
- Graph data
- Indicator variables
- Feature binning
- Interaction variables
- Mathematical transforms
- Model specific featurizations
- Feature orthogonality
- Domain specific featurizations
- Feature slicing
- Kaggle Winners solutions

#### Miscellaneous Topics

- Calibration of Models:Need for calibration
- Productionization and deployment of Machine Learning Models
- Calibration Plots.
- Platt’s Calibration/Scaling.
- Isotonic Regression
- Code Samples
- Modeling in the presence of outliers: RANSAC
- Productionizing models
- Retraining models periodically.
- A/B testing.
- Data Science Life cycle
- VC dimension

#### Case Study 1: Quora question Pair Similarity Problem

- Business/Real world problem : Problem definition
- Business objectives and constraints.
- Mapping to an ML problem : Data overview
- Mapping to an ML problem : ML problem and performance metric.
- Mapping to an ML problem : Train-test split
- EDA: Basic Statistics.
- EDA: Basic Feature Extraction
- EDA: Text Preprocessing
- EDA: Advanced Feature Extraction
- EDA: Feature analysis.
- EDA: Data Visualization: T-SNE.
- EDA: TF-IDF weighted Word2Vec featurization.
- ML Models :Loading Data
- ML Models: Random Model
- ML Models : Logistic Regression and Linear SVM
- ML Models : XGBoost
- Assignments

#### Case Study 2: Personalized Cancer Diagnosis

- Business/Real world problem : Overview
- Business objectives and constraints.
- ML problem formulation :Data
- ML problem formulation: Mapping real world to ML problem.
- ML problem formulation :Train, CV and Test data construction
- Exploratory Data Analysis:Reading data & preprocessing
- Exploratory Data Analysis:Distribution of Class-labels
- Exploratory Data Analysis: “Random” Model
- Univariate Analysis:Gene feature
- Univariate Analysis:Variation Feature
- Univariate Analysis:Text feature
- Machine Learning Models:Data preparation
- Baseline Model: Naive Bayes
- K-Nearest Neighbors Classification
- Logistic Regression with class balancing
- Logistic Regression without class balancing
- Linear-SVM.
- Random-Forest with one-hot encoded features
- Random-Forest with response-coded features
- Stacking Classifier
- Majority Voting classifier
- Assignments.

#### Case Study 3:Facebook Friend Recommendation using Graph Mining

- Problem definition.
- Overview of Graphs: node/vertex, edge/link, directed-edge, path.
- Data format & Limitations.
- Mapping to a supervised classification problem.
- Business constraints & Metrics.
- EDA:Basic Stats
- EDA:Follower and following stats.
- EDA:Binary Classification Task
- EDA:Train and test split.
- Feature engineering on Graphs:Jaccard & Cosine Similarities
- PageRank
- Shortest Path
- Connected-components
- Adar Index
- Kartz Centrality
- HITS Score
- SVD
- Weight features
- Modeling

#### Case study 4:Taxi demand prediction in New York City

- Business/Real world problem Overview
- Objectives and Constraints
- Mapping to ML problem :Data
- Mapping to ML problem :dask dataframes
- Mapping to ML problem :Fields/Features.
- Mapping to ML problem :Time series forecasting/Regression
- Mapping to ML problem :Performance metrics
- Data Cleaning :Latitude and Longitude data
- Data Cleaning :Trip Duration.
- Data Cleaning :Speed.
- Data Cleaning :Distance.
- Data Cleaning :Fare
- Data Cleaning :Remove all outliers/erroneous points
- Data Preparation:Clustering/Segmentation
- Data Preparation:Time binning
- Data Preparation:Smoothing time-series data.
- Data Preparation:Smoothing time-series data cont..
- Data Preparation: Time series and Fourier transforms.
- Ratios and previous-time-bin values
- Simple moving average
- Weighted Moving average.
- Exponential weighted moving average
- Results.
- Regression models :Train-Test split & Features
- Linear regression.
- Random Forest regression
- Xgboost Regression
- Model comparison
- Assignment.

#### Case study 5: Stackoverflow tag predictor

- Business/Real world problem
- Business objectives and constraints
- Mapping to an ML problem: Data overview
- Mapping to an ML problem:ML problem formulation.
- Mapping to an ML problem:Performance metrics.
- Hamming loss
- EDA:Data Loading
- EDA:Analysis of tags
- EDA:Data Preprocessing
- Data Modeling : Multi label Classification
- Data preparation.
- Train-Test Split
- Featurization
- Logistic regression: One VS Rest
- Sampling data and tags+Weighted models.
- Logistic regression revisited
- Why not use advanced techniques
- Assignments.

#### Case Study 6: Microsoft Malware Detection

- Business/real world problem :Problem definition
- Business/real world problem :Objectives and constraints
- Machine Learning problem mapping :Data overview.
- Machine Learning problem mapping :ML problem
- Machine Learning problem mapping :Train and test splitting
- Exploratory Data Analysis :Class distribution.
- Exploratory Data Analysis :Feature extraction from byte files
- Exploratory Data Analysis :Multivariate analysis of features from byte files
- Exploratory Data Analysis :Train-Test class distribution
- ML models – using byte files only :Random Model
- k-NN
- Logistic regression
- Random Forest and Xgboost
- ASM Files :Feature extraction & Multiprocessing.
- File-size feature
- Univariate analysis
- t-SNE analysis.
- ML models on ASM file features
- Models on all features :t-SNE
- Models on all features :RandomForest and Xgboost
- Assignments.

#### Unsupervised learning/Clustering

- What is Clustering?
- Unsupervised learning
- Applications
- Metrics for Clustering
- K-Means: Geometric intuition, Centroids
- K-Means: Mathematical formulation: Objective function
- K-Means Algorithm.
- How to initialize: K-Means++
- Failure cases/Limitations
- K-Medoids
- Determining the right K
- Code Samples
- Time and space complexity
- Assignment-10: Apply K-means, Agglomerative, DBSCAN clustering algorithms

#### Hierarchical clustering Technique

#### DBSCAN (Density based clustering) Technique

- Density based clustering
- MinPts and Eps: Density
- Core, Border and Noise points
- Density edge and Density connected points.
- DBSCAN Algorithm
- Hyper Parameters: MinPts and Eps
- Advantages and Limitations of DBSCAN
- Time and Space Complexity
- Code samples.
- Assignment-10: Apply K-means, Agglomerative, DBSCAN clustering algorithms
- Revision Questions

#### Recommender Systems and Matrix Factorization

- Problem formulation: Movie reviews
- Content based vs Collaborative Filtering
- Similarity based Algorithms
- Matrix Factorization: PCA, SVD
- Matrix Factorization: NMF
- Matrix Factorization for Collaborative filtering
- Matrix Factorization for feature engineering
- Clustering as MF
- Hyperparameter tuning
- Matrix Factorization for recommender systems: Netflix Prize Solution
- Cold Start problem
- Word vectors as MF
- Eigen-Faces
- Code example.
- Assignment-11: Apply Truncated SVD
- Revision Questions

#### Interview Questions on Recommender Systems and Matrix Factorization.

#### Case Study 7: Amazon fashion discovery engine(Content Based recommendation)

- Problem Statement: Recommend similar apparel products in e-commerce using product descriptions and Images
- Plan of action
- Amazon product advertising API
- Data folders and paths
- Overview of the data and Terminology
- Data cleaning and understanding:Missing data in various features
- Understand duplicate rows
- Remove duplicates : Part 1
- Remove duplicates: Part 2
- Text Pre-Processing: Tokenization and Stop-word removal
- Stemming
- Text based product similarity :Converting text to an n-D vector: bag of words
- Code for bag of words based product similarity
- TF-IDF: featurizing text based on word-importance
- Code for TF-IDF based product similarity
- Code for IDF based product similarity
- Text Semantics based product similarity: Word2Vec(featurizing text based on semantic similarity)
- Code for Average Word2Vec product similarity
- TF-IDF weighted Word2Vec
- Code for IDF weighted Word2Vec product similarity
- Weighted similarity using brand and color
- Code for weighted similarity
- Building a real world solution
- Deep learning based visual product similarity:ConvNets: How to featurize an image: edges, shapes, parts
- Using Keras + Tensorflow to extract features
- Visual similarity based product similarity
- Measuring goodness of our solution :A/B testing
- Exercise :Build a weighted Nearest neighbor model using Visual, Text, Brand and Color

#### Case study 8:Netflix Movie Recommendation System (Collaborative based recommendation)

- Business/Real world problem:Problem definition
- Objectives and constraints
- Mapping to an ML problem:Data overview.
- Mapping to an ML problem:ML problem formulation
- Exploratory Data Analysis:Data preprocessing
- Exploratory Data Analysis:Temporal Train-Test split.
- Exploratory Data Analysis:Preliminary data analysis.
- Exploratory Data Analysis:Sparse matrix representation
- Exploratory Data Analysis:Average ratings for various slices
- Exploratory Data Analysis:Cold start problem
- Computing Similarity matrices:User-User similarity matrix
- Computing Similarity matrices:Movie-Movie similarity
- Computing Similarity matrices:Does movie-movie similarity work?
- ML Models:Surprise library
- Overview of the modelling strategy.
- Data Sampling.
- Google drive with intermediate files
- Featurizations for regression.
- Data transformation for Surprise.
- Xgboost with 13 features
- Surprise Baseline model.
- Xgboost + 13 features +Surprise baseline model
- Surprise KNN predictors
- Matrix Factorization models using Surprise
- SVD ++ with implicit feedback
- Final models with all features and predictors.
- Comparison between various models.
- Assignments.

#### Deep Learning:Neural Networks.

- History of Neural networks and Deep Learning.
- How Biological Neurons work?
- Growth of biological neural networks
- Diagrammatic representation: Logistic Regression and Perceptron
- Multi-Layered Perceptron (MLP).
- Notation
- Training a single-neuron model.
- Training an MLP: Chain Rule
- Training an MLP:Memoization
- Backpropagation.
- Activation functions
- Vanishing Gradient problem.
- Bias-Variance tradeoff.
- Decision surfaces: Playground

#### Deep Learning: Deep Multi-layer perceptrons

- Deep Multi-layer perceptrons:1980s to 2010s
- Dropout layers & Regularization.
- Rectified Linear Units (ReLU).
- Weight initialization.
- Batch Normalization.
- Optimizers:Hill-descent analogy in 2D
- Optimizers:Hill descent in 3D and contours.
- SGD Recap
- Batch SGD with momentum.
- Nesterov Accelerated Gradient (NAG)
- Optimizers:AdaGrad
- Optimizers : Adadelta andRMSProp
- Adam
- Which algorithm to choose when?
- Gradient Checking and clipping
- Softmax and Cross-entropy for multi-class classification.
- How to train a Deep MLP?
- Auto Encoders.
- Word2Vec :CBOW
- Word2Vec: Skip-gram
- Word2Vec :Algorithmic Optimizations.

#### Deep Learning: Tensorflow and Keras.

- Tensorflow and Keras overview
- GPU vs CPU for Deep Learning.
- Google Colaboratory.
- Install TensorFlow
- Online documentation and tutorials
- Softmax Classifier on MNIST dataset.
- MLP: Initialization
- Model 1: Sigmoid activation
- Model 2: ReLU activation.
- Model 3: Batch Normalization.
- Model 4 : Dropout.
- MNIST classification in Keras.
- Hyperparameter tuning in Keras.
- Exercise: Try different MLP architectures on MNIST dataset.

#### Deep Learning: Convolutional Neural Nets.

- Biological inspiration: Visual Cortex
- Convolution:Edge Detection on images.
- Convolution:Padding and strides
- Convolution over RGB images.
- Convolutional layer.
- Max-pooling.
- CNN Training: Optimization
- Example CNN: LeNet [1998]
- ImageNet dataset.
- Data Augmentation.
- Convolution Layers in Keras
- AlexNet
- VGGNet
- Residual Network.
- Inception Network.
- What is Transfer learning.
- Code example: Cats vs Dogs.
- Code Example: MNIST dataset.
- Assignment: Try various CNN networks on MNIST dataset.

#### Deep Learning: Long Short-term memory (LSTMs)

#### Interview Questions on Deep Learning

#### Case Study 9: Human Activity Recognition

#### Case Study 10: Self Driving Car

- Self Driving Car :Problem definition.
- Datasets.
- Data understanding & Analysis :Files and folders.
- Dash-cam images and steering angles.
- Split the dataset: Train vs Test
- EDA: Steering angles
- Mean Baseline model: simple
- Deep-learning model:Deep Learning for regression: CNN, CNN+RNN
- Batch load the dataset.
- NVIDIA’s end to end CNN model.
- Train the model.
- Test and visualize the output.
- Extensions.
- Assignment.

#### Case Study 11: Music Generation using Deep-Learning

- Real-world problem
- Music representation
- Char-RNN with abc-notation :Char-RNN model
- Char-RNN with abc-notation :Data preparation.
- Char-RNN with abc-notation:Many to Many RNN ,TimeDistributed-Dense layer
- Char-RNN with abc-notation : State full RNN
- Char-RNN with abc-notation :Model architecture,Model training.
- Char-RNN with abc-notation :Music generation.
- Char-RNN with abc-notation :Generate tabla music
- MIDI music generation.
- Survey blog:

#### SQL

- Introduction to Databases
- Why SQL?
- Execution of an SQL statement.
- IMDB dataset
- Installing MySQL
- Load IMDB data.
- USE, DESCRIBE, SHOW TABLES
- SELECT
- LIMIT, OFFSET
- ORDER BY
- DISTINCT
- WHERE, Comparison operators, NULL
- Logical Operators
- Aggregate Functions: COUNT, MIN, MAX, AVG, SUM
- GROUP BY
- HAVING
- Order of keywords.
- Join and Natural Join
- Inner, Left, Right and Outer joins.
- Sub Queries/Nested Queries/Inner Queries
- DML:INSERT
- DML:UPDATE , DELETE
- DDL:CREATE TABLE
- DDL:ALTER: ADD, MODIFY, DROP
- DDL:DROP TABLE, TRUNCATE, DELETE
- Data Control Language: GRANT, REVOKE
- Learning resources

#### Interview Questions

#### Live session Videos

#### Applications of non-gaussian distributions? How to use correlations?Download Our Syllabus ( click on CURRICULUM tab to view lessons)

## Obective of Applied AI/ Machine Learning Online Course:

The AppliedAICourse attempts to teach students/course-participants some of the core ideas in machine learning, data-science and AI that would help the participants go from a real world business problem to a first cut, working and deployable AI solution to the problem. Our primary focus is to help participants build real world AI solutions using the skills they learn in this course.

This course will focus on practical knowledge more than mathematical or theoretical rigor. That doesn’t mean that we would water down the content. We will try and balance the theory and practice while giving more preference to the practical and applied aspects of AI as the course name suggests. Through the course, we will work on 20+ case studies of real world AI problems and datasets to help students grasp the practical details of building AI solutions. For each idea/algorithm in AI, we would provide examples to provide the intuition and show how the idea to used in the real world.

### Key Points:

- Validity of this course is 365 days( i.e Starts from the date of your registration to this course)
- Expert Guidance, we will try to answer your queries in atmost 24hours
- 10+ real world case studies and 5 case studies will be given as assignments to build your portfolio. please click here to view the sample portfolio
- 30+ machine learning and Deep learning algorithms will be taught in this course.
- No prerequisites– we will teach every thing from basics ( we just expect you to know basic programming)
- Python for Data science is part of the course curriculum.
- The content of this course will be dynamic(i.e lessons will be added if there is an exceptional paper published)

### Target Audience:

We are building our course content and teaching methodology to cater to the needs to students at various levels of expertise and varying background skills. This course can be taken by anyone with a working knowledge of a modern programming language like C/C++/Java/Python. We expect the average student to spend at least 5 hours a week over a 6 month period amounting to a 145+ hours of effort. More the effort, better the results. Here is a list of customers who would benefit from our course:

Undergrad (BS/BTech/BE) students in engineering and science.

- Grad(MS/MTech/ME/MCA) students in engineering and science.
- Working professionals: Software engineers, Business analysts, Product managers, Program managers, Managers, Startup teams building ML products/services.
- ML Scientists and ML engineers.

#### Click Here to Download Our Syllabus

<

### Course Features

- Lectures 733
- Quizzes 0
- Duration 140+ hours
- Skill level All levels
- Language English
- Students 5249
- Assessments Yes

## 524 Comments

I want to learn about Named Entity Recognition(NER). Please suggest some links with code samples.

Thanks,

Mehul

please go through these blogs – blog1, blog2, blog3

Ok thanks.

I’m a management student in business analytics.I don’t have much programming skills.Is this relevant for people who want to start their career as data analyst/business analyst? Are there people from this domain who enrolled in this course?

Please feel free to call us at +91 8106-920-029 or +91 6301-939-583 so that one of us can better understand your background and we can have a detailed discussion with you on this.

hello sir this is divyam raj i was looking for a data science course is this course helpful?

Yes. Irrespective of the educational background, we are able to place many students in ML Engineer/Data Scientist positions. For detailed discussion, please call us at

91-8106-920-029(Within India)or mail us atteam@appliedaicourse.comHello Sir, I was going through your tutorials of Exploratory Data Analysis. Can you please tell me where bin_edges and count in CDF are explained?.

I would say that you complete the videos serially and don’t skip in between but if you are still interested then start watching from

https://www.appliedaicourse.com/course/applied-ai-course-online/lessons/histogram-and-introduction-to-pdfprobability-density-function-1/

Can u also include Forecasting

We already have case studies ‘Taxi Demand Prediction’ on time series forecasting problem. Please go through that.

I am 29 years old. I had completed Bsc. IT and Working as an executive since last 4 years. Would you please suggest me that should this course is useful for me. If I want to make my career in machine learning. Is it possible for me to change my current profile to machine learning. Please guide me.

Please feel free to call us at +91 8106-920-029 or +91 6301-939-583 so that one of us can better understand your background and we can have a detailed discussion with you on this.

Hello sir

Please can you explain the attention and self attention in your videos.

The concept of Attention is simple if you understand what it does. It basically takes the output from all the time steps in the encoder and append it and pass it in the decoder section in something called Context Vector so that the decoder can see which encode the state of LSTM was important to decode.

Thanks for replying.

I have confusion in self attention, what does keys query and value mean and what would be the dimension of these?

I want to understand QA-net that basically uses self attention please help.

Drop us an email at team@appliedaicourse.com and we shall help you out.

Hello ,

I am btech(IT) graduate and it’s been 6 months i started working as a Quality engineer in a software company , my concern is that i have my 12th score is less than 60% , but above 70% in btech and 10th would my 12th percentage become any kind of a pebble here in job scenario

Please reply in this context

Please feel free to call us at +91 8106-920-029 or +91 6301-939-583 so that one of us can better understand your background and we can have a detailed discussion with you on this.

Is this course recommended for students directly after High School. It it equivalent to the professional degrees for career building? Happy to discuss over phone call.

Yeah sure, Please feel free to call us at +91 8106-920-029 or +91 6301-939-583 so that one of us can have a detailed discussion with you on this.

Can you also add excel tutorial like you have add for MySQL.

Sorry, currently we are not having any plans of adding those lectures on excel.

Hii team@appliedaicourse.com,

I’m running three days behind which I having trouble following the schedule, that is impacting on my study. I’m not properly following the schedule that you created.

Please help me sir

you need to be very hard on the schedule, if you are lagging on a few days, you can spend more time on weekends and align with the schedule.

i have not done any good project in my Btech final year.But i am applying for pgee i.e entrance exam for iiit hyderabad. In the application form they are asking for my project details. so in what way this course is helping me to do the project within two months of time.And i have no idea about machine learning projects. is there any project included in this course ?

Case studies, comprising of problems from various domains are part of the course.

regarding ability to complete the course in two months, this will depends on learning potential and the number of hours one can spare daily.

My system configuration is i5(5th gen), 8 Gb RAM and 4 GB Nvidea Graphic card(920M).

Can you please tell me is it good for this course?

Yes, you can go ahead with this 🙂

Plz give me total ipynb pdf drive link

We have structured the course such that whenever a concept is being covered, link for the ipython notebook(wherever there is a code explanation) is provided in the same page.

Hi,

I have basic question

What is the difference between AI, ML, and Data science ??

check this out: https://www.youtube.com/watch?v=QizsAE4fBpQ

sir how many months it would take to complete the course and what should be the prerequisite for the course?

6 months and pre-requisite is basic mathematics understanding. Everything will be taught end to end.

what if someone is weak in mathematics from the basic then should he enrolled to this course

Dont worry. All the mathematics required will be covered as part of the course in simple and easy to understand terms.

Hi. I would like to know what system requirements we require to follow the course assignments and lectures. Is it possible to run the assignments or any given code in Google Colab or Kaggle Kernels, as they provide free GPU(also TPU in Colab case). I am really interested in joining the course and would like to know clear recommended specifications that our machine should possess. Apologies if the question is repeated.

Any laptop with the below configurations is fine.

System requirements: 8G -16G RAM, with i7 processor, 4G graphics card from NVIDIA(if you want to run deep learning programs which is optional, generally 1050Ti and above is preferable).

You can use Google Colab for deep learning problems as well.

Could you further clarify about which i7 processor variant to choose exactly? If it’s no trouble. Because the other specifications are easy to choose. Thanks

Anything can be fine:

https://ark.intel.com/content/www/us/en/ark/products/series/122593/8th-generation-intel-core-i7-processors.html

I did ML course in coursera previously, Now i decided to do applied Ai course.My current qualifications is Diploma(cse) . Just suppose i have done with all the casestudies nd submit all assinments. Can i get an internship in ML or data science role in industry after completing this course?

Yes, we do help you with both internships and full-time roles as you progress through and complete the course. Please call us +91 8106-920-029 or +91 6301-939-583 to know more details.

thanks i clearly understood all things

Hi

Can you please clarity a doubt .

Will Google AutoML eliminate the need for ML specialists in future ?

Please share your views regarding this.

Thanks,

Pankaj

Audio replies:

https://soundcloud.com/applied-ai-course/automl-vs-ml-engineer

https://soundcloud.com/applied-ai-course/automated-ml

Hi,

Thanks for your reply 🙂

Hello Team,

I am Kunal Singh, from Bangalore. I have worked as a data analyst in a company for the tenure of 2.2 years. I belong form the Commerce stream as i have completed my BBM. But i do have a good knowledge of Excel and MS products coz i was used to it in my profession. As i had worked in IT company i wish to phosphor in this domain.

I am Afraid of not having any Technical educational background like B.tech or CS. As i have done some of the research and Investigation i came to know the Programming languages like C,C++ and Java are important skills to learn ML and AI. I have engaged myself on learning these stuffs as of now.

As per my above concern, would be i able to Join AAIC and learn AI and ML. I can See all the other friends here belongs from the technical background 🙁 🙁 :(.

I am keen to Learn.

Please suggest me.

Please feel free to call us at +91 8106-920-029 or +91 6301-939-583 and we can explain to you how students who have no coding background could still land a job at the end of the course.

Hi Team,

The Live sessions are they going to be recorded and saved somewhere?

Regards,

Sandeep

yes, but we strongly encourage our students to attend live sessions so that it becomes a regular habit. Our registered students would be able to access older live sessions after a significant delay and not immediately after the session.

since your course offer jobs .. so what is range of salary to the fresher , who completed your course with your assignment with your given time duration

Kindly give us a call on 91-6301-939-583 (or) 91-8106-920-029 to have further and detailed discussion on it.

The content in each of the videos are intuitive also many tough topics are elucidated clearly. I really appreciate if you could also include more Deep Learning content along with Reinforcement Learning content which is totally absent in the course.

We intentionally left out Reinforcement learning not to overwhelm the students. We have covered most of the key concepts in deep learning and would surely add more in the near future. Thank you for your feedback.

Hi Team,

Are there any practice questions available for the initial lectures up to 4.11?

You can check in our classroom, we have some assignments regarding python.

i did not get any email about joining google classroom

We have sent an email with all the guidelines, if you still didn’t get please drop an email at team@appliedaicourse.com

Hi,

I works at a MNC as Java Developer.I am a Bsc graduate in Mathematics.Can I go for this course? If yes, then how will be my IT carrer here after?

am i able to manage all this because as i said I am an full time employee in a MNC?

I will begreatful if you let me know the process and all.

Thank you.

Hi, I am a 2016 batch passout in production engineering student with 6 months of experience in manufacturing sector but now I feel like switching my profile to Machine learning. So, would this course be beneficial to me as I have Non IT background with almost 2 years of gap after B Tech. Please help me out..

Can you please add one more complete module on timeseries

Time series analysis is a statistical technique that deals with time series data, or trend analysis. Basically, Time series data means that data is in a series of particular time periods or intervals. Time series are analyzed in order to understand the underlying structure and function that produce the observations.

We also solve case study “taxi demand prediction” based on time series data.

We cover a lecture on how to generate features from time series data starting from:

https://www.appliedaicourse.com/course/applied-ai-course-online/lessons/moving-window-for-time-series-data/

Hi Team,

As i have completed by b.tech in 2016 and i have got job opportunity in 2018 working as full stack developer and i want to move to AI , how can this course helpful to me as a employee after working hours i have less time to cover this course . And i want to know is there will be any problem for me while attending interviews after completing this course where i have a gap after my graduation.

I am considering going into DS/ML/AI and this course seems to fulfill most of the requirements. Sir, can you please confirm the fee / payment structure? On the website it say one year validity for $25K. Will I have access to video/lectures after one year. I would like to have continued access to material to come back for revisions. Let me know please, I really like the course and would like to enroll.

Please call us at +91 8106-920-029 or +91 6301-939-583 so that we can better answer your question more accurately.

Hello Team,

How many assignments are included totally in this course? (including case study assignments)

Thirty assignments, as of now. We constantly change these 30 assignments on regular basis.

Sir,

I have heard that Statistics is one of the most important, if not THE most important, part of Data Science. So, does it make sense to first get a solid understanding of the core concepts of Statistics before starting with the course?? That way maybe one can get even more purchase from the lectures. Or will the lectures under the “Probability And Statistics” section of the course suffice for understanding the Statistics portion of the Data Science???

We cover all of the core and foundational concepts needed for Data science on the Probability and Statistics chapter.

What is the eligibility for the job guarantee which you provide?

Please call us on +91 8106-920-029 or +91 6301-939-583 to know more.

okay sir, one more query to ask, is this machine learning course complete or you will be adding more content to it?

The course is never complete as we constantly add, edit and remove topics based on industry needs and students feedback. We have changed a very significant amount of of the course in the last 3 months. It is a very dynamic course

What is the difference between tensorflow and tensorflow 2.0?

Which part of this course will get affected by tensorflow 2.0?

tensorflow2.0 is yet to be released and there are going to some changes: https://medium.com/tensorflow/whats-coming-in-tensorflow-2-0-d3663832e9b8. Most of the deep learning modules are covered in keras so it should not be too much of a worry.

Hi ,

It will be great if you can provide a keyboard shortcut to directly move to next video 🙂 .

Thanks,

Pankaj

Sure. Will add it as a requirement for the next update of our website. Thank you for your suggestion.

Hi Team,

Could you pls tell me if I am working on Insurance Fraud Claim.If it is being told that Random forest is among the best algorithm,

then why not to apply it directly on the Insurance dataset,instead of trying any others(Since RF is well tested algorithm)?

Pls clarify in real world scenario.

Secondly,What is SMOTE?How to apply?

Thanks

1. Why not use GBDT or SVM or another method? Why RandomForest? It is best to choose the best-performing algorithm that also satisfies other business and real-world constraints like latency, interpretability, cost of training and evaluation, availability of probabilistic class labels etc.

It is best to try out various methods which satisfy the real-world constraints and pick the one that actually performs best. It is best to avoid any favorites. If the data shows that RF is the best method for your task, then you should certainly go with it.

2. SMOTE is a method to generate synthetic and new data-points using k-NN especially when we want to tackle the class-imbalance problem through upsampling.

Refer:https://en.wikipedia.org/wiki/Oversampling_and_undersampling_in_data_analysis#SMOTE

After successfully completing the course.Am I able to create a chatbot?

Yes, you will be able to build a chatbot by the end of this course.

Hi Team,

What is the difference between Regularized Logistic regression and linear svm?

Thanks

Both objectives have the same components: the loss function and the L2 penalty. It is also true that these two models often give very similar performance.

However there is fundamental difference: Logistic Regression outputs probabilities and SVM maximizes the margin. The point of SVMs is that is ignores all but the close points (support vectors) in the loss. In that sense it “maximizes margin” quite explicitly in a way that logistic regression does not.

Hi Team, i see there is not a lot of NLP related content in the course. Can you suggest a few links with good content for NLP?

This course consists of many of the important concepts from NLP ranging from simpler concepts like Bag-of-words to state of the art RNNs for NLP. We have introduced these concepts as the course progresses. We also have many case-studies which focus on NLP applications like quora-question pair similarity, Stackoverflow tagging, Personalized cancer diagnosis etc.

audio reply: https://soundcloud.com/applied-ai-course/nlp-in-the-course-1

Links: https://blog.algorithmia.com/introduction-natural-language-processing-nlp/

https://www.analyticsvidhya.com/blog/2017/01/ultimate-guide-to-understand-implement-natural-language-processing-codes-in-python/

i am curently in second year of engineering, my cgpa is low(around 6 cgpa) , even if i cover this course with good grades , will i get offer from top notch companies. will cgpa affect my career in anyways?

Please call us at +91 8106-920-029 or +91 6301-939-583 so that one of us can better understand your background and we can have a detailed discussion with you on this.

Hi Team,

I am from a testing background but i know coding basics as i have worked on selenium with Java. I am having total 7yrs of experience in IT as a tester, but i really want to move to data science or AI. I want to change my career path , will it be ok for me to move to AI or Data science. After completing this course what will be the career prospect will i be a treated as a fresher in market. Please suggest.

Please reach out to us at +91 8106-920-029 or +91 6301-939-583 so that we can have a detailed discussion with you on this.

Hi,

how do we get requirement in which form in real world as it may be from multiple source and how do we make ready to use single format e.g. csv?

How do we present our result to client?

Thanks

1. Just like any project, you talk to all the stakeholders to understand the project requirements in detail and document them clearly.

2. Databases, Data-warehouses, CSV, text-files etc.

3. Mostly through plots and detailed writeup of the findings.

Hi Team,

Why GBDT provides better performance

and accuracy than bagging(RANDOM FOREST)?

Pls elaborate.

Thanks

GBDT does not always give better performance than random forest. but many times we will get better performance because it trains tree based on errors of before tree. random forest makes many trees with random sample data and gives the majority vote.

Why we are taking F0(x)=L(Yi-gamma) instead of F0(x)=L(Yi-Ypred) as error should be initially Yi-Ypred to start with,Rather that not being (Yi -gamma) in GBDT or XGBOOST?

Pls clarify in detail,as I am totally confused.

Thanks

gamma in base model F0 indicated predictions values only. don’t confuse with learning rate gamma, that is different.

But as in wikipedia,

F0(x)= argmin_gamma SUM_1ton{ L(Yi,gamma) }

Here, as written above gamma is a part/term of complete eqn. of F0(x).

So,how “gamma in base model F0 indicated predictions values only”?

Pls clear me as always 🙂

Thanks

please check in this video.

And one more query, I want to ask is about the case studies – some times when we understand the concept and cannot work on case study but need some more understanding of the question being asked or sometime hints, then can we contact you for the same?

We mentor you through the case studies that you would work on. More importantly, you would have seen many case-studies in the course itself and hence you would be able to take our hints and proceed. Of course, you would have to put in a reasonable and non-trivial amount of effort from your end also.

Hi Team,

I went through some of the free available videos. And I could see that you are explaining the concepts in a fairly easy manner but my concern is about the code. I could listen the faculty saying that we can download the code using notebooks but what is the meaning of each line of code? Which function is doing what? What else can we write to accomplish that same thing?

As I am from programming background, I know that we can accomplish one task with many different ways, that is why, I am a bit confused and curious to know the answer to this please.

Where can we get the answers to these queries? It would be great, if you can help me getting answers to these queries.

Thanks,

Gurmeet Kaur

Gurmeet, we explain the basics of Python in the very first chapter. We also explain each of the key functions in detail as and when we progress through the course. But, as we progress through the code, we may not go line by line as we have seen similar code earlier. If you do not understand any segment of code, feel free to put it in the comments section under the video. We would respond with a clarification. It is impossible to cover all the ways to achieve the same task as there are sometimes tens of ways to do the same using different functions and libraries. We try and provide one of the more commonly used and efficient ways. Additionally, we want our students to have the ability to navigate through ML code comfortably by the end of the course. Hence, we avoid going over code line by line in the later chapters of the course as similar concepts have been covered in earlier chapters. If there is some new concept, we would surely go over it in detail. We want our students to put in the effort to understand the code on their own especially in later chapters as they would have to do it in their jobs once they complete the course. If there is any blocker, we are always here to help.

Please call us on +91 8106-920-029 or +91 6301-939-583 to learn more about this.

HI ,

How much percentage case study weight in applied ai course ?

Thanks ,

Pankaj

Close to 40% of the content and 50% of the assignments are from case-studies.

Thanks for replying 🙂

Hi AAIC,

My friend attended interview with Accenture with 2 years Data Science Experience, below are the list of questions was asked to him:

1. Explain about yourself

2. Explain the projects that you worked for

3. What is the difference between logistic and linear

4. What is SIgmoid function

5. What is the difference between sigmoid and softmax function

6. what is validation

7. what s statification

8. what is bootstrap validation

9. What is cost function

10. Why is cost function so important in model

11. When did you apply cost function

12. What is optimizer

13.I have 10 Million records with 60 features, what is your approch to build a logistic regression

a. which validation you choose

b. which model you choose

c. Which optimizer you choose

d list the optimizers you know and explain the best one you choose for this problem

e. Where do you get the data ? is it from online or warehouse?

14.what is clustering and explain various clustering algorithms

15. Write psudo code of Density based clustering

16.Have you ever worked on SVM ? tell me the advantages

17. What is the node entropy

18. How is Rnadom forest is different from Xgboost

19. What are the most significant parameters for the xgboost

20 Why you preferr XG boost for both Liner and logistic

21. What is PCA

22. I applied PCA on 50 features and how many new feature will I get

23 WHat is Eigan vector

24. How you deploy your model into production

25. What Hadoop cluster you are using

Please give us some links or study materials if you find any topic missed in AAIC course with respect to above-listed questions.

Thanks

These are some of the easier questions in interviews. We have covered all go these concepts except (Q-25) about Hadoop in our course.

For Hadoop cluster, I guess the interviewee mentioned that he is working on Hadoop and hence he was asked which version of Hadoop and which distribution of Hadoop and probably what configuration he was using. We have also not covered (Q-1) as it changes from person to person!

Hi, can you please explain what is your answer for below questions

22. I applied PCA on 50 features and how many new feature will I get

13. I have 10 Million records with 60 features, what is your approch to build a logistic regression

a. which validation you choose

b. which model you choose

c. Which optimizer you choose

d list the optimizers you know and explain the best one you choose for this problem

e. Where do you get the data ? is it from online or warehouse?

22. You can get any number of features <=50 post PCA. If you want to pick the smallest number of features while maximizing the variance preserved through PCA, we apply the elbow method to determine the best number of features which would vary from dataset to dataset.13. First break the data randomly into train, CV and test datasets assuming there is no temporal ordering in the data-points. Next, we may not need all for the 10MM data points to build an LR model. Since LR is not a complex model, we should randomly sample the train data and build a model and measure it's performance. Increase the sample size and observe if there is any improvement in the model. At some point, increasing the sample size would not improve the model performance dramatically. Now, we can use this trained model on a much smaller sampled data than training the model on all of the 10MM points.a. simple CV or k-fold CV. b. LR+ L2-regularization. Would use L1 regularization if we want a sparse weigh-vector. c. batch-SGD. Works well in practice and is fast. d. SGD, GD, ADAM, ADAGRAD, RMSPROP, ADADELATA. This is a simple model and hence we will stick to batch-SGD. e. LR can be made to work in both cases: batch data from a warehouse or streaming data.

sir i am btech 6th semester(3rd year) cse student after completing this course will your team help me in getting an ai based job or an internship in my 4th year.

Yes, we do help you with both internships and full-time roles as you progress through and complete the course. Please call us +91 8106-920-029 or +91 6301-939-583 to know more details.

Hi Team,

Suppose I have created a model in Jupyter notebook.How do we deploy to the production enviroment in real time?

you can create an API and deploy.

Hi,

Could you pls elaborate how to create api?Pls tell me the code and steps ?

Thanks

You can check in our video.

Hi. I would like to know what system requirements we require to follow the course assignments and lectures. What if I had a laptop with i3 processor and 8 Gb Ram

Minimum Recommended PC Specifications:

RAM: 8GB or above

Processor: i5 or above

You also can use online platforms like GCP, AWS, etc for assignments.

What stipend / package can I expect under the job guarantee program as an intern or a fresher for the data science role earned with your assistance?

Please call us at +91 8106-920-029 or +91 6301-939-583 so that we understand your current role and better answer your question more accurately.

Hi Applied AI,

Course Content looks really interesting with most of the topics covered. I would love to apply for this course . I need some some information from your side.

I have around 9 years of Backend programming background. With good command on SQL. Working Knowledge of spark and python. Currently I am working in Data science team (as Backend developer.). ALso I have around 3 years of Finance domain knowledge and keen interest in Risk Management and Portfolio Management.

Given my background can this course help me in progressing my career in Machine Learning?

Please call us at +91 8106-920-029 or +91 6301-939-583 so that we understand your current role and better answer your question more accurately.

Sir, @Srikanth & team, I have gone through all the Success stories & just a visual pattern, Almost 80% of the hired ones are Freshers, & the rest have Software development experience, What about people from IT whom don’t have Software development or coding experience ?.. How does AppliedAIcourse make sure that hiring is guaranteed for such a candidate ?..

Please call us at +91 8106-920-029 or +91 6301-939-583 and we can explain to you how students who have no coding background could still land a job at the end of the course.

Sir , I am from non coding background though i understand the concept of python, i am also trained in tableau . but the real job scenario is demanding 1-3 yrs experience in relevant field….. is it sure after this training job is guareented . do u provide class room training……

Please call us at +91 8106-920-029 or +91 6301-939-583 so that we can better understand your background and discuss with you in person.

Hi,

What is likelihood,log-likelihood and maximum likelihood?

What is maximum likelihood in logistic regression?

Thanks

1. Given the observed data and a model of interest, we need to find the one Probability Density Function/Probability Mass Function (f(x|θ)), among all the probability densities that are most likely to have produced the data.

To solve this inverse problem, we define the likelihood function by reversing the roles of the data vector x and the (distribution) parameter vector θ in f(x| θ), i.e.,

L(θ;x) = f(x| θ)

In MLE, we can assume that we have a likelihood function L(θ;x), where θ is the distribution parameter vector and x is the set of observations. We are interested in finding the value of θ that maximizes the likelihood with given observations (values of x).

2.Maximum likelihood estimation (MLE) is a technique used for estimating the parameters of a given distribution, using some observed data. For example, if a population is known to follow a normal distribution but the mean and variance are unknown, MLE can be used to estimate them using a limited sample of the population, by finding particular values of the mean and variance so that the observation is the most likely result to have occurred.

3. For many applications, the natural logarithm of the likelihood function, called the log-likelihood, is more convenient to work with. This is because we are generally interested in where the likelihood reaches its maximum value: the logarithm is a strictly increasing function, so the logarithm of a function achieves its maximum value at the same points as the function itself, and hence the log-likelihood can be used in place of the likelihood in maximum likelihood estimation and related techniques.

Hi Sir,

Does this course help me in solving Kaggle problems or will be in a position to face the kaggle competitions after completion of this course

Yes, we solved many real world problems in our case studies taken from the kaggle.

Hi Team,

I have around 10 years of experience in IT industry mainly with Java and related technologies. Currently working as technical lead. Should I consider shifting to ML, AI domain now? Also, does this course covers ai as well or only ml?

This course covers the breadth of techniques across ML and AI which are currently relevant to the industry. Given your experience, it is best if you could call us and discuss this over the phone where we could understand your background and answer your questions better. Please call us at +91 8106-920-029 or +91 6301-939-583.

hi

is this course contains the concept of objects , classes, constructors and destructors ???????

No, they are not covered in this course.

Hello,

I have spent close to 18 years in IT industry performing various roles as a QA, Dev, Support and lately at Management levels. My core skill is in designing solutions for business and testing. I wish to make a career shift in AI ML domain. Let me know if this is right move at this stage of my career and if there are enough opportunities for people of my age and experience.

Tushar, It is best to discuss this over the phone so that we can better understand your background and give you critical and comprehensive feedback. Please call us on +91 8106-920-029 or +91 6301-939-583.

I don’t have knowledge on libraries like numpy, pandas….and all other libraries used in ML, is there any detailed explanation of these libraries in this course?

Yes, we will explain how to use numpy and pandas etc in the course.

Hi,

I am interested to pursue this course online. So, as a part of this, will I be installing any software(s) during the course duration? If so, then what should be the basic configuration of the laptop that I use while pursuing this course online?

For GPU you can go with 1050ti Nvidia graphic card or 1060 or 1070. Try to assemble an SSD with >=128G. Keep a minimum of 8G ram preferable 16G. The hard disk is your call.

Hi,

I have a total of 8+ years of experience in Siebel CRM data migration. I have enrolled in this course last month and am really excited and i must say i am really enjoying it. But i do have some concerns as well.

Since i have a lot of experience in CRM (particularly in migrating data from legacy systems to Siebel databases):

1. Are there opportunities in ML on CRM data in big companies?

2. I would like to do 2-3 case studies in CRM area, will Applied AI team support and mentor me in understanding the CRM scenarios and help me in completing the case studies?

3. Do companies hire 8+ years experience guy like me with fresh knowledge of ML considering the CTC expectation would be high compared to other fresh graduates who have completed the course?

Very much appreciate your reply on this. Thank you very much!

Regards,

Mohammed Arif

1. Yes, Some of the largest CRM software developers like Salesforce have invested heavily in AI and built tools like Einstein. So, CRM is certainly a very fertile area for applications of AI as it contains lots of text and customer data.

2. Certainly, as you progress through the course, we will work closely with you to build a portfolio of projects around the CRM field as it is your corer area of expertise. We will need lots of inputs from you on the domain-knowledge front as we are not experts in CRM. But, we can certainly will mentor you o the ML front.

3. We have had students who have transitioned to ML roles with your kind of experience. But, the effort needed would certainly be more as you will have to prove your ML expertise + Domain expertise as employers would expect more advanced case-studies in your portfolio. Employers do pay a premium for domain expertise, which is critical for most ML projects. I think it is too early to talk about compensation now, but we will put in our best efforts to land you in the best possible role based o your expertise.

Hi,

Is your job placement support restricted to India or you have tie-up with UAE (Dubai & Abu Dhabi) companies as well?

Thanking you in advance for your valuable reply.

We currently have the Job-Guarantee limited to India and US only. We have NOT yet expanded and built our recruiter partner network in the middle-east.

Hi,

1) I am coming from Mainframe, and from last three year I am in Big Data Support project, having working knowledge of Hive, Hadoop, sqoop, etc; and understanding of spark-Scala, python, core JAVA ( from company training and self learning). Though when ever I tried for interview, since I am 12 Work exp, question come around like architecture and not like Developer. So, what interview question company will be expecting for MI, knowing that if they investing me with good money, they need more from me, as total Work of years experience.

2) for doing assignment, do we have mentor-student and student-student forum, where we can clear our doubts ?

Audio reply: https://soundcloud.com/applied-ai-course/comment-big-data/s-89yG0

How does ridge regression solves multicollinearity?Why lasso don’t?

Yes, ridge Regression used when the data suffers from multicollinearity ( independent variables are highly correlated). In multicollinearity, even though the least squares estimates (OLS) are unbiased, their variances are large which deviates the observed value far from the true value. By adding a degree of bias to the regression estimates, ridge regression reduces the standard errors. Ridge regression solves the multicollinearity problem through shrinkage parameter λ (lambda). Have a look at the link here:

https://stats.stackexchange.com/questions/104779/why-does-ridge-regression-work-well-in-the-presence-of-multicollinearity

Hi,

I already went before through the link but couldn’t get intuation.Could you pls simplify a bit more.

And,why don’t lasso(ya I got that it sparsity very easily through your video.But,why not solves multicollinearity?).

Thanks.

If two predictors are highly correlated LASSO can end up dropping one rather arbitrarily. That’s not very good when you’re wanting to make predictions for a population where those two predictors aren’t highly correlated, & perhaps a reason for preferring ridge regression in those circumstances.

Thanks.How does multicollinearity affects Random Forest?Actually I encountered this question.

In terms of prediction, it does not affect much if the effect of multi-collinearity presents in both the train and test. However, feature importance score would be influenced, just like other methods such as linear/logistic regression.

The explanation can be simplified if we consider two collinear variables. If two strong variables are chosen for a decision tree in RF, one must be picked for each split consideration and the other would not likely be picked again since the former has already explained the output variation for both of them (*). The prediction power of that tree does not drop though, so does the RF.

However, in feature importance score, once one of the two is picked, the other variable is less important to bring any further output variation explanation since the two are correlated well. The overall importance of these two will be reduced

Hi Applied AI team,

Seaborn topic is not covered in EDA, Without basics unable to understand the code related to the seaborn package, Can anyone help me on it.

Thank you,

Regards,

bayaprakash

If you know matplotlib, Seaborn is quite straightforward to understand and implement. Have a look at this: https://www.tutorialspoint.com/seaborn/index.htm

Does this course contains –

1. OpenCV

2. Audio Recognition topics (including audio preprocessing)

1. We do cover OpenCV to the extent we need for our case-studies involving self-driving car. We donot dive deep into OpenCV as this is not a computer vision course.

2. We perform lots of time-series analysis using Fourier transforms and discuss about Spectrograms as a featurization of audio data. We have a case-study about Music generation where we touch upon some audio topics like music encodings. But, as this is not an audio-processing course, we do not dive too deep into audio-signal processing and dive deep into techniques like Mel Frequency Cepstral Coefficients (MFCC).

As part of our course, you will gain the ability to learn many concepts in CV and Audio-processing and use these concepts for ML tasks. We have had students who learned about and used MFCC to build singer detection systems. Similarly, our students were able to pick up SIFT and other feature engineering techniques in CV and apply ML models on top of them.

Hi Team,

please correct me if I am Wrong

1. C.I. is tool/technique which is using to making Rough not exact calculated estimation about data.

2. It Gives us confidence that my calculation or understanding about data is x% of range is Right rather than throwing stones in the darkness.

But do we need to take samples of data every time like it shouldn’t be mandatory activity, Right?

thanks.

Yes, C.I gives you a range of values for a random-variable (like mean height of students in a class) and a confidence level (95%) associated with it to signify our “confidence” in this range.

Without samples of X or the parameters of the distribution(like mu and sigma for Normal disb) of X, we cannot come up with a C.I.

I think this question does not belong in this comment section. Please post questions under the respective and relevant videos.

Hi,

I’m very new to ML, just covering statistics in Udemy. My 3 questions.

1. Is it possible to get lifetime access and can I get access all case studies?

2.I’m beginner, How many hours I need to spend per day to complete course and hands on?

3. Do I get certification and mentor to guide me?

Hi Selva Kumar,

1. We are currently restricting the access to one year only to ensure that students are disciplined to complete the course on time.

2. Please use this sheet to fill in your effort so that we can give you a detailed topic-wise timeline that an average student of our course has taken to complete the course. You will get a detailed timesheet with efforts and duration within a day.

3. You will get a certificate upon course completion. We have a dedicated team to help you throughout the course answering your queries and questions within 24 hrs. You will also get a mentor to help you with your own case-studies and help you prepare better for placements once you complete 50% of the assignments in the course.

Please call us at +91 8106-920-029 or +91 6301-939-583 to know more.

Hi Team,

I have one question which may seem silly:-

“Ok. Suppose I did EDA and came out with some observations and intuation about data

(eg.,after univariate,bivariate,descriptive,etc. analysis like how data is distributed,related,segmented).

Now, what we do with these observations( before fitting to sklearn model,eg. LogisticRegression().fit() )?

I mean what is the purpose of these and where it is used for(before fitting into model)?

Is its done to SELECT FEATURES(by identifying and dropping the non-useful features), ENGINEER FEATURES(derivation of new features),impute missing values and handle outliers——->””””””””””””””””””SO THAT WE CAN HAVE REFINED DATASET TO FEED INTO THE models””””””””””””””””””””””

Is the quoted sentence is the sole purpose to get and collect intuation by EDA?

Thanks

Yes, you are right. There are more reasons for it as well:

1. Let’s take for example you’d want to apply gaussian naive Bayes. Now the assumption is that the features are supposed to be Gaussian distributed and this requires us to explore our data to determine the distribution of our features.

2. You can also perform use correlation tests to explore which features are collinear(this creates a problem of interpretability of weight vectors in logistic/linear regression).

There are many such reasons why EDA is necessary and on a case, to case basis, you recognize what kind of EDA you’d perform.

Please tell this course covers all the aspects required for Cloudera certification (CCA, CCP) or not?

This is NOT a course on Cloudera certification for Cloudera Big data platform. We cover ML and AI in this course which are not part of CCP or CCA.

Not exactly. Since you already have some experience in networking, we work with you on networking related ML case-studies inn the course once you finnish most of the contents. This will help you build on top of the domain expertise you have in networking. That will help you land a job in companies that are looking for folks in the inntersection of network-security and ML. But, 2 years is not a big deal. So, you can also look for ML roles outside of networking domain also. It is very common for people to transition across jobs in thier career and many recuiters do udnerstand that. While the recuiritng tream might consider you close to a fresher for ML roles, your networking exprience will also be countned and given some weightage.

Hi,

Where can I find free videos?

Please register and then log into AppliedAICourse using your Google/FB/Linkedin Credentials.

You can access our free videos here which are mmrked with a green-colored eye icon here:

https://www.appliedaicourse.com/course/applied-ai-course-online/lessons/how-to-learn-from-appliedaicourse/

Please let us know if you encounter anny probelms.

Hi Team,

I was just going through your free videos.

I was watching this video of EDA with Iris dataset, unfortunately I was not able to plot a 2D scatter plot with Petal Length and Petal Width instead I could plot it with Sepal Length and Sepal Width.

Can you please guide me through this code so that I can identify the problem as I am looking forward to enroll for this course.

Kindly go through these blogs:

1. https://machinelearningmastery.com/machine-learning-in-python-step-by-step/

2. https://medium.com/codebagng/basic-analysis-of-the-iris-data-set-using-python-2995618a6342

Hi Team,

What is word2vec(Embedding)?Are we covering this?Where it is used in NLP?

Thanks

Word embedding is one of the most popular representation of document vocabulary. Loosely speaking, they are vector representations of a particular word. Word2Vec is one of the most popular technique to learn word embeddings using shallow neural network. Yes, we discussed the concept of word2vec in detail and how exactly it works in NLP.

Thanks Team.

Could you pls give the link for “Word Embedding” in the course content.

And Thumbs up for your effort.Well Done.

It has been covered in the multi-layered perceptron section of the course. Check here

Hello Sir,

I found your content interesting but unfortunately, I downloaded the iPython Notebooks, I am wondering why I can’t use Jupyter Notebook as a tool it always get open in the browser.

I would like to know if I can really use it as a tool or I always have to use it in the browser, if so how will I save my notes and whenever I download this Ipython notes it does not open in the format of Jupyter Notebook, infact it gives me multiple option such as MS Word, Notepad,etc.Can I Use Spyder for the same? Or is it that the code of Spyder can’t run in Jupyter or vice-versa.

Please help me with this as I am new to programming language and also looking forward to join this course.

The Jupyter Notebook App is a server-client application that allows editing and running notebook documents via a web browser. So you need to start the application with “jupyter notebook” on the terminal. If you’d like to use it with an editor, you have to convert it to .py file. Have a look at this and this. Your notebooks can be saved and edited like regular python scripts, just that you’ll be required to open it by launching it on your local system.

Thanks Sir,

I would just like to know whether I can use Spyder 3.6 for the same and another challenge is to install packages from Command Prompt

Whenever I use pip3 it states pip3 is not recognized.

1. It might either be that you haven’t added python to your system variables: https://superuser.com/questions/143119/how-do-i-add-python-to-the-windows-path or just try pip instead and see if it works.

2. You can launch spyder with the anaconda navigator.

I’m B.E graduate in Mechanical Engineering with 57% with a decent knowledge in programming and I’m also familiar with Machine Learning in a intermediatory level, can i still learn this course and get placed?

Please call us at +91 8106-920-029 or +91 6301-939-583 so that one of us can better understand your background and suggest you if this course is correct for you or not.

can you give me the answer that you got from appliedaiteam for this question

hii there,

Which engineering domain is eligible to take this course.As I am from EEE and have knowledge of programming language of c and c++ .Am I eligible to take this course.

Yes, you can take this course, before that please check out the free videos. please mail us at team@appliedaicourse.com for further queries.

Hello.

I’m B.E graduate with 55% and not good in programs, can i still learn this course and get placed?

To be honest, you got to work very hard to land any job given your prior record and lack of programming knowledge. You got to skill up and be able to showcase your skills which got to be very good. Please call us at +91 8106-920-029 (or) +91 6301-939-583 to discuss whether this course is correct for you or not.

Did I miss the Time series in index or it is not in the course yet?

we have couple of case studies Taxi demand prediction and HAR you will learn the concepts and featurization of time series data in those case studies.

HI,

I have attended multiple online courses on ML, basically all were hard core theoretical, will this course benefit me ?

and are you guys planning to Add some NLP stuff in this course ?

Hi Shubham, We strongly recommend you go through our sample/free videos which you can access by logging in to our course so that you will get a good idea of how we balance theory and practice of ML. We focus a lot on helping you build a portfolio of projects which help you showcase your skills to potential recruiters. We are very focused on helping our students transition to ML careers through our job-guarantee or money-back guarantee program.

This course consists of many of the important concepts from NLP ranging from simpler concepts like Bag-of-words to state of the art RNNs for NLP. We have introduced these concepts as the course progresses. We also have many case-studies which focus on NLP applications like quora-question pair similarity, Stackoverflow tagging, Personalized cancer diagnosis etc.

oh ok.. regarding Job gaurantee,

just wanted an assurance that it wont be like setting up an interview for a non reliable company or a company paying in peanuts ?

Please look at some of our success stories for a list of companies where our students have been placed. Our students have been placed in a wide spectrum of companies from Fortune 500 to smaller ones based on our student’s calibre, interests and skills. Our student’s salaries also have varied from Rs 5 Lakhs to Rs 30 Lakhs per annum based on their level of expertise and portfolio they build at the end of the course.