_4SortMovieForUser.py

import pandas as pd
import numpy as np
import matrix_lowRankMatrixFactorization


class predictMovies_forUser:
    # Load data from csv
    movie_names = pd.read_csv("SampleMovieData/movies.csv")
    raw_dataset_df = pd.read_csv("SampleMovieData/ratings.csv")
    raw_dataset_df = raw_dataset_df.iloc[:500, :3]  # Specify the no. of rows to get from file(100 for less data)

    def create_HTML_CSV(self, data, name):
        data_df = pd.DataFrame(data)
        html = data_df.to_html(na_rep="")
        with open('_4' + str(name) + '.html', 'w') as f:
            f.write(html)
        data.to_csv('_4' + str(name) + '.csv')

    def getAllPredicted_Ratings(self):
        # Convert the list of user ratings into matrix =>to build review matrix
        ratings_df = pd.pivot_table(self.raw_dataset_df, index="userId", columns="movieId", aggfunc=np.max)

        # Given Data
        self.create_HTML_CSV(ratings_df, "Given")

        # to replace nan with 0
        ratings_df = ratings_df.fillna(value=0)

        # Matrix factorization: to find User and Movie attribute matirx
        U, M = matrix_lowRankMatrixFactorization.matrix_factorization(np.array(ratings_df))
        Output = np.round_((np.dot(U, M)), 2)
        # adding row and column in final_Output
        col = ratings_df.columns.values
        print(ratings_df.index.values)
        new_col = []
        for i in col:
            new_col.append(i[1])
        predicted_Ratings = pd.DataFrame(Output, index=ratings_df.index.values, columns=new_col)

        # Predicted Ratings
        self.create_HTML_CSV(predicted_Ratings, "PredictedForAll")
        print("Predicted Ratings for All::")
        print(predicted_Ratings)
        #print("------------------------------------")

        self.predictForUser(predicted_Ratings)

    # ----------------------------NEW Code (particular User Recommendation)-----------------



    def predictForUser(self, predicted_Ratings):
        print("Enter a UserId to get Top 10 Recommendations (user b/w 1 to 10 )")
        userId_toSearch = int(input())
        print("Movies already reviewed by user:" + str(userId_toSearch))

        Already_Reviewed_byUser = self.raw_dataset_df[self.raw_dataset_df['userId'] == userId_toSearch]
        print(Already_Reviewed_byUser)
        PredictedRating_forGivenUser = (predicted_Ratings.T[userId_toSearch])
        All = PredictedRating_forGivenUser

        y = (np.setdiff1d(All.keys(), np.array(Already_Reviewed_byUser['movieId'])))
        zz = (self.movie_names.loc[np.sum(self.movie_names['movieId'].values == pd.DataFrame(y).values, axis=0) == 1])
        zz = np.array(zz)
        rating = np.array(All.loc[np.in1d(np.array(All.keys()), y)])
        zz = np.append(zz, rating.reshape(rating.shape[0], 1), axis=1)
        zz = pd.DataFrame(zz)
        final_Data = zz.sort_values(3, ascending=False)
        index = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25]
        final_Data = pd.DataFrame(final_Data,index=index)
        finals = final_Data.sort_values(3, ascending=False)
        finals.columns = ["MovieId", "Movie Name", "Genre", "Predicted Rating for User " + str(userId_toSearch)]
        print(finals)

        self.create_HTML_CSV(finals, "Recommendation_User")


if (__name__ == "__main__"):
    predictMovies_forUser().getAllPredicted_Ratings()