Home Courses Netflix Movie recommendation system Intuitive understanding of bias-variance.

Intuitive understanding of bias-variance.

Instructor: Applied AI Course Duration: 6 mins

This content is restricted. Please Login

Bias-Variance tradeoff

Best and worst cases for an algorithm

Real world problem: Predict rating given product reviews on Amazon

1.1

Dataset overview: Amazon Fine Food reviews(EDA)

23 min

1.2

Data Cleaning: Deduplication

15 min

1.3

Why convert text to a vector?

14 min

1.4

Bag of Words (BoW)

18 min

1.5

Text Preprocessing: Stemming, Stop-word removal, Tokenization, Lemmatization.

15 min

1.6

uni-gram, bi-gram, n-grams.

9 min

1.7

tf-idf (term frequency- inverse document frequency)

22 min

1.8

Why use log in IDF?

14 min

1.9

Word2Vec.

16 min

1.10

Avg-Word2Vec, tf-idf weighted Word2Vec

9 min

1.11

Bag of Words( Code Sample)

19 min

1.12

Text Preprocessing( Code Sample)

11 min

1.13

Bi-Grams and n-grams (Code Sample)

5 min

1.14

TF-IDF (Code Sample)

6 min

1.15

Word2Vec (Code Sample)

12 min

1.16

Avg-Word2Vec and TFIDF-Word2Vec (Code Sample)

2 min

1.17

Exercise: t-SNE visualization of Amazon reviews with polarity based color-coding

6 min

Classification And Regression Models: K-Nearest Neighbors

2.1

How “Classification” works?

10 min

2.2

Data matrix notation

7 min

2.3

Classification vs Regression (examples)

6 min

2.4

K-Nearest Neighbors Geometric intuition with a toy example

11 min

2.5

Failure cases of KNN

7 min

2.6

Distance measures: Euclidean(L2) , Manhattan(L1), Minkowski, Hamming

20 min

2.7

Cosine Distance & Cosine Similarity

19 min

2.8

How to measure the effectiveness of k-NN?

16 min

2.9

Test/Evaluation time and space complexity

12 min

2.10

KNN Limitations

9 min

2.11

Decision surface for K-NN as K changes

23 min

2.12

Overfitting and Underfitting

12 min

2.13

Need for Cross validation

22 min

2.14

K-fold cross validation

17 min

2.15

Visualizing train, validation and test datasets

13 min

2.16

How to determine overfitting and underfitting?

19 min

2.17

Time based splitting

41 min

2.18

k-NN for regression

5 min

2.19

Weighted k-NN

8 min

2.20

Voronoi diagram

4 min

2.21

Binary search tree

16 min

2.22

How to build a kd-tree

17 min

2.23

Find nearest neighbours using kd-tree

13 min

2.24

Limitations of Kd tree

9 min

2.25

Extensions

3 min

2.26

Hashing vs LSH

10 min

2.27

LSH for cosine similarity

40 min

2.28

LSH for euclidean distance

13 min

2.29

Probabilistic class label

8 min

2.30

Code Sample:Decision boundary

23 min

2.31

Code Sample:Cross Validation

13 min

2.32

Exercise: Apply k-NN on Amazon reviews dataset

5 min

Classification algorithms in various situations

3.1

Introduction

5 min

3.2

Imbalanced vs balanced dataset

23 min

3.3

Multi-class classification

12 min

3.4

k-NN, given a distance or similarity matrix

9 min

3.5

Train and test set differences

22 min

3.6

Impact of outliers

7 min

3.7

Local outlier Factor (Simple solution :Mean distance to Knn)

13 min

3.8

K-Distance(A),N(A)

4 min

3.9

Reachability-Distance(A,B)

8 min

3.10

Local reachability-density(A)

9 min

3.11

Local outlier Factor(A)

21 min

3.12

Impact of Scale & Column standardization

12 min

3.13

Interpretability

12 min

3.14

Feature Importance and Forward Feature selection

22 min

3.15

Handling categorical and numerical features

24 min

3.16

Handling missing values by imputation

21 min

3.17

Curse of dimensionality

27 min

3.18

Bias-Variance tradeoff

24 min

3.19

Intuitive understanding of bias-variance.

6 min

3.20

Best and worst cases for an algorithm

6 min

Performance measurement of models

4.1

Accuracy

15 min

4.2

Confusion matrix, TPR, FPR, FNR, TNR

25 min

4.3

Precision and recall

10 min

4.4

Receiver Operating Characteristic Curve (ROC) curve and AUC

19 min

4.5

Log-loss

12 min

4.6

R-Squared

14 min

4.7

Median absolute deviation (MAD)

5 min

4.8

Distribution of Errors

7 min

Logistic Regression

5.1

Geometric intuition of Logistic Regression

31 min

5.2

Sigmoid function: Squashing

37 min

5.3

Mathematical formulation of Objective function

24 min

5.4

Weight vector

11 min

5.5

L2 Regularization: Overfitting and Underfitting

26 min

5.6

L1 regularization and sparsity

11 min

5.7

Probabilistic Interpretation: Gaussian Naive Bayes

19 min

5.8

Loss function interpretation

24 min

5.9

Hyperparameter search: Grid Search and Random Search

16 min

5.10

Column Standardization

5 min

5.11

Feature importance and Model interpretability

14 min

5.12

Collinearity of features

14 min

5.13

Test/Run time space and time complexity

10 min

5.14

Real world cases

11 min

5.15

Non-linearly separable data & feature engineering

28 min

5.16

Code sample: Logistic regression, GridSearchCV, RandomSearchCV

28 min

5.17

Exercise: Apply Logistic regression to Amazon reviews dataset.

6 min

5.18

Extensions to Logistic Regression: Generalized linear models

9 min

Linear Regression

6.1

Geometric intuition

13 min

6.2

Mathematical formulation

14 min

6.3

Real world cases

9 min

6.4

Code sample for Linear Regression

13 min

Solving Optimization Problems

7.1

Differentiation

29 min

7.2

Online differentiation tools

8 min

7.3

Maxima and Minima

12 min

7.4

Vector calculus: Grad

10 min

7.5

Gradient descent: geometric intuition

19 min

7.6

Learning rate

8 min

7.7

Gradient descent for linear regression

8 min

7.8

SGD algorithm

9 min

7.9

Constrained Optimization & PCA

14 min

7.10

Logistic regression formulation revisited

6 min

7.11

Why L1 regularization creates sparsity?

17 min

7.12

Exercise: Implement SGD for linear regression

6 min