Netflix-Data-Analysis

This ia a detailed data analysis for popular streaming service Netflix. A web-app which can be used to get recommendations for a series/movie, the app recommends a list of media according to list of entered choices of movies/series in your preferred language.

Table of contents

1. Objective
2. Getting data
3. Preprocessing and feature selection
4. Model Selection

---

Objective

The objective of this problem is to analyse users watch history and recommend them shows based on the genre that are most watched by the users.

Getting data

Here the users watch history is analysed and based on the genres that are watched by the users a table is prepared. Here is the sample table data showing genres, actors, and tags that were most popular among the shows that a particular user has watched.

Preprocessing and feature selection

Here the data is converted into the format that can be used by neural network. Here each observation in dataset mentioned above is a TV-show and season combination. For example, Season 1 of “13 Reasons Why”.^[1] Here genres, tags, and season number as categorical variables, and episode length as a numeric variable.
Then a corelation matrix is created that removes the highly co-related features like removing the overfitting that can be caused due to the corelation of crime and murder genre.
Here is the code block for the syntax of corelation matrix

columns = np.full((X_corr.shape[0],), True, dtype=bool) 
#Keep only columns with correlation less than 0.9 
for i in range(X_corr.shape[0]): 
    for j in range(i+1, X_corr.shape[0]): 
        if X_corr.iloc[i,j] >= 0.9: 
            if columns[j]: columns[j] = False 
selected_X = X[X.columns[columns]] 

Model selection

Then the best model to run on is selected keeping in mind number of nodes, the regularization coefficient, and the dropout value. For that grid search is used.

---

Libraries used

• Pandas
• Numpy
• Matplotlib
• Plotly

---

Installation and how to run

• For installing the above mentioned libraries write the below code snippet in terminal.
  pip install <libraryname>
• Then import the libraries.
• This project can be run on Google Colab.

---

Author

• Github: https://github.com/heema-1989/Netflix-Data-Analysis/edit/main/README.md
• Email: heemag2002@gmail.com

---

References

[1] https://towardsdatascience.com/using-neural-nets-to-recommend-shows-on-netflix-d4fbecffe0b0