Recommendation Systems

  • By
  • April 30, 2022
  • Machine Learning

Recommendation Systems – 

A few days back, my 16 years old brother came to me and asked me, 

“You study all this Artificial Intelligence, Machine Learning. Can you make those replicas of  humans, those robots which can think on their own, is it really possible? why can’t you make a  robot for me, will he do my homework?”  

Here I was struggling to classify bananas with oranges, he wants me to make a robot for him. Self Respect Crushed in a second!!!  

After regaining my self-respect, I thought, If I will say, “No This is not possible”, then Self Respect  Crushing episodes will become daily soaps for me, so I thought to make him understand Machine  Learning Intuition.  

For Free, Demo classes Call:  7507414653

Registration Link: Click Here!

I made him sit and tried to make him understand, data cleaning, data preprocessing, model training,  evaluations, and whatnot.  

After almost half an hour, after listening to everything, he just stood up and said, “just say that you  can’t make a robot for me, don’t bore me”  

Self Realisation moment for me, I am the worst teacher for sure!!!  

I was thinking about this situation, how can I make him understand ML/AI in a way that excites  him, make him understand the beauty of Machine Learning rather than just waiting for a robot to do  his homework, which I can’t make right now for sure. 

I challenged myself to write a blog that can give intuition or that can give the flavour of Machine  Learning to my brother.  

Not only to my brother, but this blog could also be the starting point for any kid who is excited  about Artificial Intelligence, Machine Learning. This article is for all the people who believe in  technology, who believes AI is the future but does not know what exactly is Artificial Intelligence.  

Before starting, I want to request all the readers that I am writing this blog as a challenge to myself,  the challenge is very simple, “Anybody who have at least passed High School will be able to  understand the flavour of AI after this blog, does not matter whether you are in Sales, Hr, Finance or  you are just a student, you will see AI as an opportunity rather than some complex difficult  science.”  

If after reading, you do not get the intuition of Machine Learning/AI, if this field does not excites you, please drop a small message, I will try to update this blog as much I can.  

For Free, Demo classes Call:  7507414653

Registration Link: Click Here!

Let’s get started –

After the major insult episode from my brother, i understood one thing very clearly that i need to  change my approach. I have to think from his end rather than thinking from my end. 

I am gonna use simple technique called Reverse Engineering.  

Lets talk about a very famous company called NETFLIX…  

Let’s talk about How Netflix is using Machine Learning to solve their problems.  

Netflix is simply a OTT platform where we can watch some of the best movies and shows of the  world, Now Netflix earns from its subscription model, so more the Netflix Subscribers, more the  Revenue. Now Netflix wants you to get excited for the shows , Netflix want you to stay as more as  possible on their Website.  

Netflix wants their users to stay on their website as much as possible, They want to  recommend you all the shows/movies which you can’t escape. 

 

 

How they do that??  

They use something called Recommendation system…  

Let’s say we have two types of Recommendation System:-  

  1. USER USER SIMILARITY  
  2. ITEM ITEM SIMILARITY  

 

USER USER SIMILARITY  –

 

 

So these kind of recommendation is very very common in industry these days as whether you see an  e-commerce company like Amazon, a music company like Spotify, a OTT platform like Netflix etc.  

For Free, Demo classes Call:  7507414653

Registration Link: Click Here!

It simply says, “ SIMILAR PEOPLE SEE SIMILAR KIND OF CONTENT, SIMILAR  PEOPLE LISTEN SIMILAR KIND OF MUSIC, SIMILAR PEOPLE BUYS SIMILAR  KIND OF THINGS”

I am sure you had encounter a friend song list very similar to yours, this happens lot of times in real  world. Many times people like same kind of food, same kind of music, same kind of shows and this  becomes the backbone for this approach.  

Let me take an example, 

Let’s take 3 friends, Nishesh, Rohit and Mrinalini, Nishesh and Mrinalini both are on Netflix from  last few years, Rohit has joined recently.  

Nishesh likes 

  1. Stranger Things(Thriller) 
  2. Sacred Games(Indian Thriller) 
  3. Little Things(Indian Rom-com) 
  4. The Game Changer( Health documentary) 

Mrinalini Likes  

  1. Spiderman(Far from Home) 
  2. Daredevil(Marvel Show) 
  3. Friends(Famous Rom-Com) 
  4. Stranger Things(Thriller) 
  5. Sacred Games 

Now Rohit has recently joined the Netflix subscription so he just watched  

  1. Stranger Things  

Now what do you think, what Netflix should recommend to Rohit, Now because Mrinalini and  Nishesh Both have watched Sacred Games after Stranger things , it is most likely that Rohit will  also like it, so this is USER USER SIMILARITY.  

ITEM-ITEM SIMILARITY –

 

Now the other type of recommendation system is Item Item similarity, Here, we explore the  relationship between the pair of items (the user who bought Y, also bought Z). We find the missing  rating with the help of the ratings given to the other items by the user.  

For Free, Demo classes Call:  7507414653

Registration Link: Click Here!

Now connecting this to the above example, Lets say Rohit saw Sacred Games and after watching he  also watched Little Things, So Now if Netflix want to recommend something to Rohit using using  Item Item similarity, then because Rohit saw a ROM-COM there are more chances he may also like  watching more Romantic Comedies so Netflix also recommended “FRIENDS” to him so thats what  is ITEM ITEM similarity.  

Thanks for Reading… 

 

Author:-

Nishesh Gogia

© Copyright 2021 | Sevenmentor Pvt Ltd.

Call the Trainer and Book your free demo Class  Call now!!!


| SevenMentor Pvt Ltd.

 

Submit Comment

Your email address will not be published.

*
*