GitHub

Used FastAPI framework to build API endpoints .
Implemented Docker to containerize the application.

FastAPI model code 🚀

                    
                        import pandas as pd
                        # import requests


                        def ml_recommendation_model(movie: str, rating: int):

                            # # loading the data to use correlation algorithm
                            # movies = pd.read_csv('./data/movies.csv')
                            # ratings = pd.read_csv('./data/ratings.csv')

                            # # data manipulation & data cleaning
                            # ratings = pd.merge(movies, ratings).drop(['genres','timestamp'], axis=1)
                            # user_ratings = ratings.pivot_table(index=['userId'], columns=['title'], values='rating')
                            # user_ratings = user_ratings.dropna(thresh=10, axis=1).fillna(0)
                            # item_similarity_df = user_ratings.corr(method='pearson')

                            ## To create .pkl file from Pandas Dataframe
                            # item_similarity_df.to_pickle('item_similarity_df.pkl')


                            # Directly Loading pre-created .pkl file/model for more efficient web-page loading
                            item_similarity_df = pd.read_pickle('./data/item_similarity_df.pkl')

                            # Function to get similar movies
                            def get_similar_movies(movie_name, user_rating):
                                similar_score = item_similarity_df[movie_name] * (user_rating - 2.5)
                                similar_score = similar_score.sort_values(ascending= False)
                                return similar_score

                            # Function to check & remove, if the recommended movies is already seen
                            def check_seen(movie, seen_movie):
                                if movie == seen_movie:
                                    return True
                                else:
                                    return False

                            # Empty Dataframe
                            similar_movies = pd.DataFrame() 
                            
                            # Empty similar movies list
                            similar_movie_list = []

                            try:
                                
                                # Getting similar movies
                                similar_movies = similar_movies._append(get_similar_movies(movie, rating), ignore_index= True)

                                all_recommend = similar_movies.sum().sort_values(ascending= False)


                                # Logic to check & remove, if the recommended movies is already seen
                                check_title = movie
                                i = 0

                                for recommended_movie, score in all_recommend.items():
                                    if not check_seen(recommended_movie, check_title):
                                        similar_movie_list.append(recommended_movie)
                                    else:
                                        pass

                                    i = i + 1
                                    if i >= 30:
                                        break

                                # return the recommended/similar movies
                                return similar_movie_list

                            except:
                                # return empty list
                                return similar_movie_list