from implicit.als import AlternatingLeastSquares
from implicit.approximate_als import FaissAlternatingLeastSquares
import pandas as pd
from tqdm import tqdm, trange
from scipy.sparse import coo_matrix, csr_matrix
import numpy as np
import deepdish as dd
import pickle
import os
from dask import dataframe
import pandas as pd
import gc
os.environ["MKL_NUM_THREADS"] = "1"raw_data = pd.read_csv("/data/ECommAI/round2_a/3Tuple/raw_data.csv", names=["user_index", "item_index", "ratings"])
user_id = raw_data["user_index"].values
item_id = raw_data["item_index"].values
ratings = raw_data["ratings"].values

behavior_map = {"clk": 1.0 ,"collect":2.0 ,"cart" : 3.0 ,"buy" : 4.0}

item_user_data = csr_matrix((ratings, (item_id, user_id)), shape=(max(item_id) + 1, max(user_id) + 1))
print("Done")
# initialize a model
model = AlternatingLeastSquares(factors=64, use_gpu=True, iterations=100, calculate_training_loss=True)
# train model
model.fit(item_user_data)
dd.io.save("/data/ECommAI/round1_b/user_factors.h5", model.user_factors, compression=False)
dd.io.save("/data/ECommAI/round1_b/item_factors.h5", model.item_factors, compression=False)

import deepdish as dd
import tensorflow as tf
import numpy as np
from tqdm import trangetf.enable_eager_execution()
user_factors = dd.io.load("/data/ECommAI/round1_b/user_factors.h5")
item_factors = dd.io.load("/data/ECommAI/round1_b/item_factors.h5")

# recommend_result = Nonedef make_recommendation(index, scale):item_tensor = tf.convert_to_tensor(item_factors)user_tensor = tf.convert_to_tensor(user_factors[index:index + scale, :])idx0 = tf.shape(user_tensor)[0]if idx0 != 0:recommendation = tf.tensordot(user_tensor, item_tensor, axes=[1, 1])recommendation_k = tf.math.top_k(recommendation, k=50)return recommendation_k[1]else:return None

scale = 200
recommend_result = make_recommendation(0, scale).numpy()user_count = user_factors.shape[0]
for i in trange(1, user_count // scale + 1):recommend_result = np.concatenate((recommend_result, make_recommendation(i * scale, scale).numpy()), axis=0)dd.io.save("/data/ECommAI/round1_b/ALS_recommendations.h5", recommend_result, compression=None)