Your IP : 216.73.216.52

Current Path : /proc/thread-self/root/home/deltalab/PMS/recommendations/user_profiling/
Upload File :
Current File : //proc/thread-self/root/home/deltalab/PMS/recommendations/user_profiling/app_testMultiProcessing.py

from components.ProfilingUsers import ProfilingUsers
from components.ProfileBased_RS import ProfileBased_RS

from _library.visual_utils import visualize_user_profile
from _library.data_loader import load_indacoProducts
from _library.mongodb_utils import simplified_SKUs
from _library.io_toolkit import write_profiles, saveComputationalTimes

import multiprocessing as mp

import os

GENERATE_RECOMMENDATIONS = True

if __name__ == '__main__':
    
    # Read data
    indacoProducts_df, indacoOrders_df, indacoCategories = load_indacoProducts(visualize_aggregated_territories = True)
    
    # ----------------------------------- PRE-PROCESSING ----------------------------------------------
    
    # [PRE-PROCESSING A] Simplify SKUs
    indacoProducts_df, product_names, sku_mapping = simplified_SKUs(indacoProducts_df, product_identifier = 'SKU')
    
    # [PRE-PROCESSING B] Add the simplified SKUs to the orders
    reversed_skuMapping = {indaco_sku: simplified_sku for simplified_sku, indaco_sku in sku_mapping.items()}
    indacoOrders_df.insert(loc = 4, column = 'SKU', value = indacoOrders_df['sku'].apply(
        lambda indaco_sku: reversed_skuMapping[indaco_sku]))
    indacoOrders_df.rename(columns = {'sku': 'indaco_sku'}, inplace = True)
    
    # Enhanced the order with information of the products
    duplicate_columns = ['Product Type', 'indaco_sku', 'Title']
    enhancedOrders = indacoOrders_df.merge(indacoProducts_df.drop(columns = duplicate_columns), how = 'left', on ='SKU')
    enhancedOrders = enhancedOrders.dropna(subset = ['SKU'])
    
    # Generate profile for the users users
    profiler = ProfilingUsers(orders = enhancedOrders)
    userProfiles = profiler.mine_orders()
    
    # --------------------------- RECOMMENDER SYSTEM ----------------------------
    if GENERATE_RECOMMENDATIONS:
        
        verbose = False
        
        # Initialize
        userBased_recomSys = ProfileBased_RS(
            platformProducts = indacoProducts_df, 
            platfromOrders = enhancedOrders)
        
        # Multiprocessing
        num_cpu = mp.cpu_count()
        pool = mp.Pool(num_cpu)
        print("Num CPU:", num_cpu)
        
        
        # Generate recommendations
        userRecommendations_byUsers, computationalTime_byUser = dict(), dict()
        
        # Multiprocessing
        worker_manager = mp.Manager()
        queue = worker_manager.Queue()
        sys_processes = []
        
        args = []
        for user_id, user_profile in userProfiles.items():
            args.append((queue, user_id, user_profile, verbose))
            #process = mp.Process(
            #    target = userBased_recomSys.userWise_recommendations_multiProcess, 
            #    args = (queue, user_id, user_profile, verbose))
            #process.start()
            #sys_processes.append(process)
            
        with pool:
            workers = pool.apply_async(userBased_recomSys.userWise_recommendations_multiProcess, args)
            
            while(not queue.empty()):
                user_recommendations, user_computationalTime = queue.get_nowait() 
                userRecommendations_byUsers.update(user_recommendations)
                computationalTime_byUser.update(user_computationalTime)
        

        # Save the results
        #for p in sys_processes:
        #    user_recommendations, user_computationalTime = queue.get() 
            
        #    userRecommendations_byUsers.update(user_recommendations)
        #    computationalTime_byUser.update(user_computationalTime)
        
        # Wait all processes
            
        # Save computational times
        saveComputationalTimes(computationalTime_byUser)
    # ----------------------------------------------------------------------------

    # Visualize the user profiles
    print("\nUSER PROFILES:")
    for user_id, user_profile in userProfiles.items():
        
        # Merge the user profile with the recommendended products 
        if GENERATE_RECOMMENDATIONS:
            user_recommendations = userRecommendations_byUsers[user_id]
            user_profile.update(user_recommendations)
        
        # Visualize profile
        visualize_user_profile(user_id, user_profile)
        
    # Write the user profiles
    write_profiles(userProfiles, sku_mapping)