Tweet
Tweet

Slide 1

Slide 1 text

A Practical Guide to SVD

Slide 2

Slide 2 text

Have you ever wondered why you binge-watch N E T F L I X ?

Slide 3

Slide 3 text

How did Spotify find out you secret love for Taylor Swift?

Slide 4

Slide 4 text

Recommendations

Slide 5

Slide 5 text

Who am I? ● ● ●

Slide 6

Slide 6 text

What is a Recommendation Engine?

Slide 7

Slide 7 text

Why are recommendation engines important? 35% of Amazon’s revenue are generated by its recommendation engine 75% of users select movie based on Netflix’s recommendations

Slide 8

Slide 8 text

Collaborative Filtering

Slide 9

Slide 9 text

Collaborative Filtering: Users, Products, Ratings Users Ratings Products

Slide 10

Slide 10 text

What is Collaborative Filtering?

Slide 11

Slide 11 text

What is Collaborative Filtering?

Slide 12

Slide 12 text

What is Collaborative Filtering?

Slide 13

Slide 13 text

What is Collaborative Filtering?

Slide 14

Slide 14 text

Singular Value Decomposition

Slide 15

Slide 15 text

Collaborative Filtering through SVD

Slide 16

Slide 16 text

What are features?

Slide 17

Slide 17 text

What are latent features?

Slide 18

Slide 18 text

SVD Visualized User Product Matrix

Slide 19

Slide 19 text

SVD Visualized

Slide 20

Slide 20 text

Products matrix Prediction( ) = -2.0 * -0.5 = 3.52 Predicting new scores performing dot product + 6.3 * 0.4

Slide 21

Slide 21 text

Comparing based on Latent Features LF1 Users matrix

Slide 22

Slide 22 text

Demo