You are on page 1of 17

Rizwan Khalid

01-235181-047

BSIT-7A

Group Member:

Asjad Ali

01-235181-007

Project Phase_1

Movie recommendation system


Recommended engine generally in three types content Based recommended engine, collaborative
recommender engine and hybrid recommended engine

Expected Submission
With the help of this particular data set we have to build a recommended engine. And our
recommended engine will return maximum 10 movies name if an user search for a particular movie.

Evaluation
Recommended engine must return 5 movie names and maximum it can return 10 movie names if an
user search for a particular movie. This recommender engine should not give suggestion in between 1 to
4 and 6 to 10 it have to return 5 movie names for 10 movie names.

Import Dataset:
Identifying types of titles:

Dropped Unnecessary columns:

We have dropped duration, description etc. These columns will not affect our data set at any case so we
dropped them.

Split numerical and categorical data:

Finding missing values from a dataset:


We have found director, cast, country and rating columns which has missing values.

Update missing values with most frequent item’s technique:

As you can see there is no null value in a dataset.

We have categorical columns in dataset so we have to encode them for preprocessing.

Encoding:

Find and relace method OR label Encoding:


All categorical columns are converted in to numerical:
Preprocessing techniques:

1. Normalization and Rescaling:

2. Standardization:

Data Visualization:
Histogram:

Boxplot:

Line Chart:
Bar Plot:
Scatter Plot:
Project Phase_2

Content based recommendation Engine:


There are cases when the user is new on a platform and we end up having no prior information on
the user. In such scenarios, we recommend similar items based on comments, feedbacks, reviews,
description of the items with which the user interacts.

Models/ Algorithms like TF-IDF score, word2vec are used to capture the similarity in Content
Based RS.

The goal of this project is to develop a content-based recommendation engine for movies and TV
shows on Netflix. I will compare two different methods:

1. Using cast, director, country, rating and genres as features.


2. Using the words in the movie/TV show descriptions as features.

Using cast, director, country, rating and genres as features:

For Movies:
For TV:
Search function for Movies and TV:
Recommended result for movies:
Second Method
Using the words in the movie/TV show descriptions as features

Filtering movies using description:


Search using description:

Recommended Result using description:

You might also like