recommender.py

from data import CITIES, BUSINESSES, USERS, REVIEWS, TIPS, CHECKINS
from helpers import *

import random
import data
from data import *

def recommend(user_id=None, business_id=None, city=None, n=10):
    """
    Returns n recommendations as a list of dicts.
    Optionally takes in a user_id, business_id and/or city.
    A recommendation is a dictionary in the form of:
        {
            business_id:str
            stars:str
            name:str
            city:str
            adress:str
        }
    """
    if user_id is not None:
        # create dataframe containing business id and all categories
        businessdict= {}
        for city in CITIES:
            for business in BUSINESSES[city]:
                businessdict[business["business_id"]] = business["categories"]
        df_categories = pd.DataFrame.from_dict(businessdict, orient= "index", columns=['categories'])
        df_categories['business_id'] = df_categories.index

        # extract categories and compute similarity based of categories
        extracted_categories = extract_categories(df_categories)
        pivoted_data = pivot_ratings(extracted_categories)
        sim = create_similarity_matrix_categories(pivoted_data)

        # create dataframe containing reviews
        for city in CITIES:
            reviews = REVIEWS[city]
            df_rev= pd.DataFrame.from_dict(json_normalize(reviews), orient='columns')
        # delete duplicates, and keep most recent ones
        df_review = df_rev.drop_duplicates(subset=["business_id", "user_id"], keep="last").reset_index()[["business_id", "stars", "user_id"]]
        ratings_utility = create_utility_matrix(df_review)

        # predict ratings
        predicted = predict_ratings(sim, ratings_utility, df_review, 0)
        sort_predicted = predicted.sort_values(by='predicted rating', ascending=False)
        for_user = sort_predicted[sort_predicted['user_id'] == user_id][:10]

        for city in CITIES:
            count = 0
            for review in REVIEWS[city]:
                if review['user_id'] == user_id:
                    count += 1
        if count < 10:
            city = random.choice(CITIES)
            return random.sample(BUSINESSES[city], n)
        else:
            return [get_business(city, i) for i in for_user['business_id']]

        

    else:
        city = random.choice(CITIES)
        return random.sample(BUSINESSES[city], n)