trackmate_model_final.py

# -*- coding: utf-8 -*-
"""TrackMate_Model_Final.ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/14Di2qGv_RcnpteM3iCMYm1eMmBctif2P
"""

pip install tensorflow_recommenders

from google.colab import drive
# df = drive.mount('/content/drive/MyDrive/amazon.csv')

# Commented out IPython magic to ensure Python compatibility.
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.metrics.pairwise import cosine_similarity
from sklearn.model_selection import train_test_split
from typing import Dict, Text
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
import tensorflow_recommenders as tfrs
from tensorflow.keras.layers.experimental.preprocessing import StringLookup
from tensorflow.keras.layers import Embedding, Input, Dense
from tensorflow.keras.models import Model
from google.colab import files

import warnings; warnings.simplefilter('ignore')
# %matplotlib inline

"""# Load Data"""

df = pd.read_csv('TrackMate Dataset.csv', on_bad_lines = 'skip')

df_new = df.copy()
df_new.tail(7)

jumlah_baris = len(df_new)
print(f"Jumlah baris dalam dataset: {jumlah_baris}")

# Data produk baru yang ingin ditambahkan

data_produk_baru_1 = [
    {
    'product_id': 'B112X5ADO8',
    'product_name': 'Greenbean Coffee',
    'category': 'Grocery and Gourmet Food',
    'discounted_price': '₹80,000',
    'actual_price': '₹80,000',
    'discount_percentage': '0%',
    'rating': 4.9,
    'rating_count': '7',
    'about_product': 'Cerry kopi yang sudah mengalami proses pasca panen dan siap panggang',
    'user_id': "AGSGSRTEZBQY64WO2HKQTV7TWFSA,AEYD5HVYAJ23CR6PTWOOIKUOIDHA,AFRMNW6TDHDZBP2UHF2K3MEAEYUA,AHICHCW6EC3BNV2IDAEAJPBG4HZQ,AGWFKE7RNP6EVC4JFLFSL76EEVVQ,AGEOQQHGNELZNEUKJAJUA7NTPBLA,AFS3QBSOMCE2FAZFUYZ3NBFQDLMQ,AGJYG6ZWCWD74WNE6Y37XZ2VUSMA",
    'user_name': "Birendra ku Dash,Aditya Gupta,Abdulla A N,Deepak,Gowtham,Rakesh,Pawan Kumar,Prabhat Raj Pathak",
    'review_id': "A1B2C3D4E5F6G7H8, I9J1K2L3M4N5O6, P7Q8R9S1T2U3V4, W5X6Y7Z8A9B1C2, D3E4F5G6H7I8J9, K1L2M3N4O5P6Q7, R8S9T1U2V3W4X5, Y6Z7A8B9C1D2E3",
    'review_title': "Rasa autentik dan khas, bahan-bahan segar terasa, penyajian cantik dan instagramable, menu ramah vegetarian, tempat yang bersih dan nyaman, suasana tenang, pilihan sehat dan bergizi, layanan ramah.",
    'review_content': 'Revolutionary, high-quality, elegant design, affordable, user-friendly, functional, fast delivery, highly recommended for any product!',
    'img_link': 'https://drive.google.com/open?id=1mCbKdwvUu9lZV-uC18Ak9_oAACyy18YH',
    'product_link': 'https://www.instagram.com/agroniaga_idn/'
},
    {
    'product_id': 'B613M6MSA8',
    'product_name': 'MORINGA TEA',
    'category': 'Grocery and Gourmet Food',
    'discounted_price': '₹40,000',
    'actual_price': '₹40,000',
    'discount_percentage': '0%',
    'rating': 4.9,
    'rating_count': '7',
    'about_product': "Produk berbahan dasar daun kelor yang bentuk kemasan teabag tinggal seduh.",
    'user_id': "AGSGSRTEZBQY64WO2HKQTV7TWFSA,AEYD5HVYAJ23CR6PTWOOIKUOIDHA,AFRMNW6TDHDZBP2UHF2K3MEAEYUA,AHICHCW6EC3BNV2IDAEAJPBG4HZQ,AGWFKE7RNP6EVC4JFLFSL76EEVVQ,AGEOQQHGNELZNEUKJAJUA7NTPBLA,AFS3QBSOMCE2FAZFUYZ3NBFQDLMQ,AGJYG6ZWCWD74WNE6Y37XZ2VUSMA",
    'user_name': "Birendra ku Dash,Aditya Gupta,Abdulla A N,Deepak,Gowtham,Rakesh,Pawan Kumar,Prabhat Raj Pathak",
    'review_id': "AGSGSRTEZBQY64WO2HKQTV7TWFSA,AEYD5HVYAJ23CR6PTWOOIKUOIDHA,AFRMNW6TDHDZBP2UHF2K3MEAEYUA,AHICHCW6EC3BNV2IDAEAJPBG4HZQ,AGWFKE7RNP6EVC4JFLFSL76EEVVQ,AGEOQQHGNELZNEUKJAJUA7NTPBLA,AFS3QBSOMCE2FAZFUYZ3NBFQDLMQ,AGJYG6ZWCWD74WNE6Y37XZ2VUSMA",
    'review_title': "Makanan sehat dan organik, pilihan vegan yang baik, desain interior yang Instagramable, suasana yang tenang dan ramah lingkungan, menu salad yang kreatif, harga terjangkau untuk makanan sehat, pelayanan yang ramah dan informatif.",
    'review_content': 'Revolutionary, high-quality, elegant design, affordable, user-friendly, functional, fast delivery, highly recommended for any product!',
    'img_link': 'https://drive.google.com/open?id=18osvUd5nE_Uu21d6c3aqi7ccPbT6dxOQ',
    'product_link': 'https://www.tokopedia.com/archive-newedenmoringa-1645404339/moringa-tea-new-eden-20gr?extParam=src%3Dshop%26whid%3D9669096'
},
    {
         'product_id': 'B404F7FMA1',
    'product_name': 'TEMPEKITA TEMPE SEGAR',
    'category': 'Grocery and Gourmet Food',
    'discounted_price': '₹17,000',
    'actual_price': '₹17,000',
    'discount_percentage': '0%',
    'rating': 4.9,
    'rating_count': '7',
    'about_product': "Tempe segar adalah produk yang berasal dari kacang kedelai yang difermentasi menggunakan ragi tempe.",
    'user_id': "AGSGSRTEZBQY64WO2HKQTV7TWFSA,AEYD5HVYAJ23CR6PTWOOIKUOIDHA,AFRMNW6TDHDZBP2UHF2K3MEAEYUA,AHICHCW6EC3BNV2IDAEAJPBG4HZQ,AGWFKE7RNP6EVC4JFLFSL76EEVVQ,AGEOQQHGNELZNEUKJAJUA7NTPBLA,AFS3QBSOMCE2FAZFUYZ3NBFQDLMQ,AGJYG6ZWCWD74WNE6Y37XZ2VUSMA",
    'user_name': "Birendra ku Dash,Aditya Gupta,Abdulla A N,Deepak,Gowtham,Rakesh,Pawan Kumar,Prabhat Raj Pathak",
    'review_id': "M8N9O1P2Q3R4S5T6U7V8W9, X1Y2Z3A4B5C6D7E8F9G1H2I3J4K5, L6M7N8O9P1Q2R3S4T5U6V7W8X9, Y1Z2A3B4C5D6E7F8G9H1I2J3K4L5, M6N7O8P9Q1R2S3T4U5V6W7X8Y9, Z1A2B3C4D5E6F7G8H9I1J2K3L4, M5N6O7P8Q9R1S2T3U4V5W6, X7Y8Z9A1B2C3D4E5F6G7",
    'review_title': "Makanan sehat dan organik, pilihan vegan yang baik, desain interior yang Instagramable, suasana yang tenang dan ramah lingkungan, menu salad yang kreatif, harga terjangkau untuk makanan sehat, pelayanan yang ramah dan informatif.",
    'review_content': 'Revolutionary, high-quality, elegant design, affordable, user-friendly, functional, fast delivery, highly recommended for any product!',
    'img_link': 'https://drive.google.com/open?id=1fv6EFhTNGeAuZgMapiOODILB3um7LgOX',
    'product_link': 'https://shopee.co.id/-Tempekita.id-Tempe-Segar-i.419555041.8543489708?xptdk=8ec5c78a-ef7b-4600-b5ab-8652b22055eb'
    },
    {
        'product_id': 'B772O6JWR8',
    'product_name': 'ZAKET',
    'category': 'Home&Kitchen',
    'discounted_price': '₹28,000',
    'actual_price': '₹28,000',
    'discount_percentage': '0%',
    'rating': 4.9,
    'rating_count': '7',
    'about_product': "Terbuat dari limbah sawit",
    'user_id': "AGSGSRTEZBQY64WO2HKQTV7TWFSA,AEYD5HVYAJ23CR6PTWOOIKUOIDHA,AFRMNW6TDHDZBP2UHF2K3MEAEYUA,AHICHCW6EC3BNV2IDAEAJPBG4HZQ,AGWFKE7RNP6EVC4JFLFSL76EEVVQ,AGEOQQHGNELZNEUKJAJUA7NTPBLA,AFS3QBSOMCE2FAZFUYZ3NBFQDLMQ,AGJYG6ZWCWD74WNE6Y37XZ2VUSMA",
    'user_name': "Birendra ku Dash,Aditya Gupta,Abdulla A N,Deepak,Gowtham,Rakesh,Pawan Kumar,Prabhat Raj Pathak",
    'review_id': "O8P9Q1R2S3T4U5V6W7X8Y9, Z1A2B3C4D5E6F7G8H9I1J2K3, L4M5N6O7P8Q9R1S2T3U4V5W6, X7Y8Z9A1B2C3D4E5F6G7H8, I9J1K2L3M4N5O6P7Q8R9S1T2, U3V4W5X6Y7Z8A9B1C2D3E4F5, G6H7I8J9K1L2M3N4O5P6, Q7R8S9T1U2V3W4X5Y6Z7",
    'review_title': "Arang dari limbah ini adalah solusi ramah lingkungan yang brilian, Mengubah limbah menjadi sumber energi yang berguna, ini membantu mengurangi jejak karbon, Proses daur ulang limbah menjadi arang sangat inovatif dan membantu menjaga lingkungan, Saya sangat mendukung produk-produk yang memprioritaskan keberlanjutan, arang dari limbah ini adalah langkah positif menuju masa depan yang lebih hijau dan berkelanjutan.",
    'review_content': 'Revolutionary, high-quality, elegant design, affordable, user-friendly, functional, fast delivery, highly recommended for any product!',
    'img_link': 'https://drive.google.com/open?id=1JTMePUNF4XMWe8IW5VW5et513nESlgWB',
    'product_link': 'https://drive.google.com/open?id=1JTMePUNF4XMWe8IW5VW5et513nESlgWB'
    },
    {
         'product_id': 'B103G6MHZ9',
    'product_name': 'Mutiara Beras Glukomanan Porang',
    'category': 'Grocery and Gourmet Food',
    'discounted_price': '₹135,000',
    'actual_price': '₹135,000',
    'discount_percentage': '0%',
    'rating': 4.9,
    'rating_count': '7',
    'about_product': "Nasi tinggi serat rendah karbohidrat rendah kalori dan rendah gula",
    'user_id': "AGSGSRTEZBQY64WO2HKQTV7TWFSA,AEYD5HVYAJ23CR6PTWOOIKUOIDHA,AFRMNW6TDHDZBP2UHF2K3MEAEYUA,AHICHCW6EC3BNV2IDAEAJPBG4HZQ,AGWFKE7RNP6EVC4JFLFSL76EEVVQ,AGEOQQHGNELZNEUKJAJUA7NTPBLA,AFS3QBSOMCE2FAZFUYZ3NBFQDLMQ,AGJYG6ZWCWD74WNE6Y37XZ2VUSMA",
    'user_name': "Birendra ku Dash,Aditya Gupta,Abdulla A N,Deepak,Gowtham,Rakesh,Pawan Kumar,Prabhat Raj Pathak",
    'review_id': "A8B9C1D2E3F4G5H6I7J8, K9L1M2N3O4P5Q6R7S8T9U1V2W3, X4Y5Z6A7B8C9D1E2F3G4H5I6J7, K8L9M1N2O3P4Q5R6S7T8U9V1W2, X3Y4Z5A6B7C8D9E1F2G3H4I5, J6K7L8M9N1O2P3Q4R5S6T7, U8V9W1X2Y3Z4A5B6C7D8, E9F1G2H3I4J5K6L7M8N9O1P2",
    'review_title': "Makanan sehat dan organik, pilihan vegan yang baik, desain interior yang Instagramable, suasana yang tenang dan ramah lingkungan, menu salad yang kreatif, harga terjangkau untuk makanan sehat, pelayanan yang ramah dan informatif.",
    'review_content': 'Revolutionary, high-quality, elegant design, affordable, user-friendly, functional, fast delivery, highly recommended for any product!',
    'img_link': 'https://drive.google.com/open?id=1xUJLdhbxytbRK9u1IdTXN9h3jETHAJVi',
    'product_link': 'https://dapurporang.com/'
    },
      {   'product_id': 'B307H9TVZ8',
    'product_name': 'Nasi Jagung Instan Loyangku',
    'category': 'Grocery and Gourmet Food',
    'discounted_price': '₹15,000',
    'actual_price': '₹15,000',
    'discount_percentage': '0%',
    'rating': 4.9,
    'rating_count': '7',
    'about_product': "Nasi Jagung Instan Loyangku terbuat dari jagung pilihan dari petani jagung Banjarnegara yang diolah secara tradisioanal oleh masyarakat Desa Pucungbedug. Yuk hidup sehat bersama Nasi Jagung Instan Loyangku.",
    'user_id': "AGSGSRTEZBQY64WO2HKQTV7TWFSA,AEYD5HVYAJ23CR6PTWOOIKUOIDHA,AFRMNW6TDHDZBP2UHF2K3MEAEYUA,AHICHCW6EC3BNV2IDAEAJPBG4HZQ,AGWFKE7RNP6EVC4JFLFSL76EEVVQ,AGEOQQHGNELZNEUKJAJUA7NTPBLA,AFS3QBSOMCE2FAZFUYZ3NBFQDLMQ,AGJYG6ZWCWD74WNE6Y37XZ2VUSMA",
    'user_name': "Birendra ku Dash,Aditya Gupta,Abdulla A N,Deepak,Gowtham,Rakesh,Pawan Kumar,Prabhat Raj Pathak",
    'review_id': "Q3R4S5T6U7V8W9X1Y2Z3A4, B5C6D7E8F9G1H2I3J4K5L6M7N8O9, P1Q2R3S4T5U6V7W8X9Y1Z2A3B4C5D6E7, F8G9H1I2J3K4L5M6N7O8P9Q1R2S3T4, U5V6W7X8Y9Z1A2B3C4D5E6F7G8H9, I1J2K3L4M5N6O7P8Q9R1S2T3U4V5, W6X7Y8Z9A1B2C3D4E5F6G7H8, I9J1K2L3M4N5O6P7Q8R9S1T2",
    'review_title': "Makanan sehat dan organik, pilihan vegan yang baik, desain interior yang Instagramable, suasana yang tenang dan ramah lingkungan, menu salad yang kreatif, harga terjangkau untuk makanan sehat, pelayanan yang ramah dan informatif.",
    'review_content': 'Revolutionary, high-quality, elegant design, affordable, user-friendly, functional, fast delivery, highly recommended for any product!',
    'img_link': 'https://drive.google.com/open?id=1auE4DYUhpOjOSn-P_Fk7akmAHzNoCj7e',
    'product_link': 'https://gubug-eva.business.site/'
    },
     {    'product_id': 'B397P6QHS8',
    'product_name': 'Sambal Baby Cumi',
    'category': 'Grocery and Gourmet Food',
    'discounted_price': '₹40,000',
    'actual_price': '₹40,000',
    'discount_percentage': '0%',
    'rating': 4.9,
    'rating_count': '7',
    'about_product': 'Sambal yang terbuat dari baby cumi asin yang berasal dari daerah desa Gerokgak kec. Gerokgak Kab. Buleleng beserta aneka cabai dan rempah-rempah bumbu segar tanpa pengawet yang diolah dengan cara dimasak sehingga memiliki rasa yang pedas, gurih dan nikmat. Cocok jika dimakan bersama nasi panas.',
    'user_id': "AGSGSRTEZBQY64WO2HKQTV7TWFSA,AEYD5HVYAJ23CR6PTWOOIKUOIDHA,AFRMNW6TDHDZBP2UHF2K3MEAEYUA,AHICHCW6EC3BNV2IDAEAJPBG4HZQ,AGWFKE7RNP6EVC4JFLFSL76EEVVQ,AGEOQQHGNELZNEUKJAJUA7NTPBLA,AFS3QBSOMCE2FAZFUYZ3NBFQDLMQ,AGJYG6ZWCWD74WNE6Y37XZ2VUSMA",
    'user_name': "Birendra ku Dash,Aditya Gupta,Abdulla A N,Deepak,Gowtham,Rakesh,Pawan Kumar,Prabhat Raj Pathak",
    'review_id': "U3V4W5X6Y7Z8A9B1C2D3E4F5, G6H7I8J9K1L2M3N4O5P6Q7R8S9, T1U2V3W4X5Y6Z7A8B9C1D2E3F4G5, H6I7J8K9L1M2N3O4P5Q6R7S8T9U1, V2W3X4Y5Z6A7B8C9D1E2F3G4H5, I6J7K8L9M1N2O3P4,D3E4F5G6H7I8J9, K1L2M3N4O5P6Q7",
    'review_title': "Makanan sehat dan organik, pilihan vegan yang baik, desain interior yang Instagramable, suasana yang tenang dan ramah lingkungan, menu salad yang kreatif, harga terjangkau untuk makanan sehat, pelayanan yang ramah dan informatif.",
    'review_content': 'Revolutionary, high-quality, elegant design, affordable, user-friendly, functional, fast delivery, highly recommended for any product!',
    'img_link': 'https://www.instagram.com/sambal_mamo/',
    'product_link': 'https://shopee.co.id/sambal_mamo-i.997481767.21685181932?sp_atk=3e60a7ec-5e55-431a-9527-c1ee7d2dca32&xptdk=3e60a7ec-5e55-431a-9527-c1ee7d2dca32'
    }
    ]

df_new = df_new.append(data_produk_baru_1, ignore_index=True)
df_new.to_csv('TrackMate Dataset.csv', index=False)

jumlah_baris = len(df_new)
print(f"Jumlah baris dalam dataset: {jumlah_baris}")

df_new.tail(7)

"""# Data Cleaning"""

# Memeriksa bentuk data
df_new.shape

"""## Checking Missing Value"""

df_new.isna().sum()

df_new.dropna(inplace=True)
df_new.isna().sum()

df_new.shape

"""## Checking Duplicates"""

# Memeriksa duplikasi data

def check_duplicates(dataframe):
    return dataframe.duplicated().sum()

print(check_duplicates(df_new))

"""## Changing Data Types"""

# Memeriksa Tipe Data

def check_data_types(dataframe):
    return dataframe.dtypes

print(check_data_types(df_new))

# Memisahkan kolom kategori

cat_split = df_new['category'].str.split('|', expand=True)
cat_split.isnull().sum()

# mengganti nama kolom

cat_split = cat_split.rename(columns={0:'Main category', 1:'Sub category'})

# mengganti kolom lama dengan yang baru

df_new['category'] = cat_split['Main category']

df_new.tail(7)

# mengganti tipe data harga diskon, harga asli dan persen diskon

df_new['discounted_price'] = df_new['discounted_price'].astype(str).str.replace('₹', '').str.replace(',', '').astype(float)
df_new['actual_price'] = df_new['actual_price'].astype(str).str.replace('₹', '').str.replace(',', '').astype(float)
df_new['discount_percentage'] = df_new['discount_percentage'].astype(str).str.replace('%','').astype(float)/100

# Menghapus data yang tidak sesuai

count = df_new['rating'].str.contains('\|').sum()
print(f"Total rows with '|' in the 'rating' column': {count}")
df_new = df_new[df_new['rating'].apply(lambda x: '|' not in str(x))]
count = df_new['rating'].str.contains('\|').sum()
print(f"Total rows with '|' in the 'rating' column': {count}")

df_new['rating'] = df_new['rating'].astype(str).str.replace(',', '').astype(float)
df_new['rating_count'] = df_new['rating_count'].astype(str).str.replace(',', '').astype(float)

print(check_data_types(df_new))

"""# Preparing the Data"""

dataset = df_new.copy()
dataset = dataset.drop(columns='product_name')
dataset = dataset.drop(columns='category')
dataset = dataset.drop(columns='discounted_price')
dataset = dataset.drop(columns='actual_price')
dataset = dataset.drop(columns='discount_percentage')
dataset = dataset.drop(columns='about_product')
dataset = dataset.drop(columns='user_name')
dataset = dataset.drop(columns='review_id')
dataset = dataset.drop(columns='review_title')
dataset = dataset.drop(columns='review_content')
dataset = dataset.drop(columns='img_link')
dataset = dataset.drop(columns='product_link')
dataset = dataset.drop(columns='rating_count')

dataset.tail(7)

dataset.isna().sum()

# Memeriksa jumlah kolom dan baris
rows, columns = dataset.shape
print("No of rows: ", rows)
print("No of columns: ", columns)

dataset.dtypes

# mencari nilai maksimal dan minimal dari rating

def find_min_max_rating():
    print('The minimum rating is: %d' %(dataset['rating'].min()))
    print('The maximum rating is: %d' %(dataset['rating'].max()))

find_min_max_rating()

# Number of unique user id and product id in the data

print('Number of unique USERS in Raw data = ', dataset['user_id'].nunique())
print('Number of unique ITEMS in Raw data = ', dataset['product_id'].nunique())

# Encode user_id
user_lookup = tf.keras.layers.StringLookup(
    vocabulary=list(dataset['user_id'].unique()),  # List of unique user_ids
    mask_token=None,
    num_oov_indices=0,
    output_mode='int',  # Output integers
    name='user_lookup'
)

# Transform user_id
dataset['encoded_user_id'] = user_lookup(dataset['user_id'])

dataset.head()

# Encode product_id
product_lookup = tf.keras.layers.StringLookup(
    vocabulary=list(dataset['product_id'].unique()),  # List of unique user_ids
    mask_token=None,
    num_oov_indices=0,
    output_mode='int',  # Output integers
    name='product_lookup'
)

# Transform product_id
dataset['encoded_product_id'] = product_lookup(dataset['product_id'])

dataset.info()

dataset.tail(7)

"""# EDA"""

# Top 10 user berdasarkan rating
most_rated = dataset.groupby('user_id').size().sort_values(ascending=False)[:10]
most_rated

# Visualisasi dengan countplot untuk kolom kategoris
plt.figure(figsize=(10, 6))
sns.countplot(x='product_id', data=dataset)
plt.title('Countplot of Product IDs')
plt.show()

# Visualisasi dengan histogram untuk kolom numerik
plt.figure(figsize=(10, 6))
sns.histplot(dataset['rating'], bins=20, kde=True)
plt.title('Histogram of rating')
plt.show()

"""# MOdeling"""

new_data = dataset.copy()

new_data = new_data.drop('user_id', axis=1)

new_data.tail(7)

new_data.info()

# Memisahkan fitur dan target
X_product = new_data[['encoded_product_id']]
X_rating = new_data[['rating']]
y = new_data['rating']

X_prod_train, X_prod_test, X_rating_train, X_rating_test, y_train, y_test = train_test_split(
    X_product, X_rating, y, test_size=0.2, random_state=42)

# Normalisasi fitur menggunakan StandardScaler
from sklearn.preprocessing import StandardScaler
scaler = StandardScaler()
X_prod_train_scaled = scaler.fit_transform(X_prod_train)
X_prod_test_scaled = scaler.transform(X_prod_test)

# Membangun model regresi linier dengan TensorFlow
from tensorflow.keras import regularizers

model_rating = tf.keras.Sequential([
    tf.keras.layers.Input(shape=(1,), name='input_product'),
    tf.keras.layers.Dense(128, activation='relu'),
    tf.keras.layers.Dense(1, activation='linear', name='output_rating')
])

# Menentukan optimizer, fungsi loss, dan metrik evaluasi
model_rating.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=0.001), loss='mean_absolute_error', metrics=['mae'])

callback = tf.keras.callbacks.EarlyStopping(monitor='val_loss', patience=3)
history = model_rating.fit(X_prod_train_scaled, X_rating_train, validation_data=(X_prod_test_scaled, X_rating_test), epochs=10)

# Evaluasi performa model rating pada set pengujian
from sklearn.metrics import mean_absolute_error

rating_predictions = model_rating.predict(X_prod_test_scaled)
mae_rating = mean_absolute_error(X_rating_test, rating_predictions)
print(f'Mean Absolute Error (Rating): {mae_rating}')

# Menampilkan plot loss dan validation
plt.plot(history.history['loss'], label='Training Loss')
plt.plot(history.history['val_loss'], label='Validation Loss')
plt.title('Training and Validation Loss')
plt.xlabel('Epoch')
plt.ylabel('Loss')
plt.legend()
plt.show()

predictions = model_rating.predict(X_prod_test_scaled)

def recommend_products(encoded_user_id, dataset, top_k_rating=5):
    # Filter dataset berdasarkan user_id dan encoded_user_id
    user_data = dataset[(dataset['encoded_user_id'] == encoded_user_id)].drop_duplicates(subset='user_id')

    # Inisialisasi DataFrame untuk hasil rekomendasi
    recommended_products = pd.DataFrame(columns=['product_id', 'product_name', 'rating_x', 'product_link'])

    # Ambil daftar kategori unik
    unique_categories = dataset['category'].unique()

    # Shuffle urutan kategori
    shuffled_categories = np.random.permutation(unique_categories)

    # Inisialisasi DataFrame sebelum loop dimulai
    recommended_products = pd.DataFrame(columns=['product_id', 'product_name', 'rating_x', 'product_link'])

    # Loop melalui setiap kategori
    for category in shuffled_categories:
        # Filter dataset berdasarkan kategori
        category_data = dataset[dataset['category'] == category]

        # Ambil sampel acak dari produk dalam kategori
        random_products = category_data.sample(n=min(top_k_rating, len(category_data))).reset_index(drop=True)

        # Gabungkan hasil rekomendasi untuk kategori saat ini ke dalam DataFrame utama
        recommended_products = pd.concat([recommended_products, random_products])

    # Urutkan hasil rekomendasi berdasarkan rating tertinggi
    recommended_products = recommended_products.sort_values(by='rating_x', ascending=False).reset_index(drop=True).head(top_k_rating)


    return user_data[['user_name']], random_products[['product_id', 'product_name', 'rating_x', 'product_link']]

# Contoh penggunaan
# Gantilah 'user_id', 'encoded_user_id', dan 'dataset' dengan data sesungguhnya
encoded_user_id_example = 3
dataset_example = pd.merge(df_new, dataset, on=['product_id', 'user_id'], how='inner')

recommendations_user, top_5_recommendations = recommend_products(encoded_user_id_example, dataset_example)

print(f"\nRekomendasi untuk {(recommendations_user).to_string(index=False)} ! :" +"\n"+ f"\n{top_5_recommendations}")

# Menyimpan model ke dalam format H5
model_rating.save('/content/TrackMate_Model.h5')

files.download('/content/TrackMate_Model.h5')

import os
import tensorflow as tf

model = tf.keras.models.load_model('TrackMate_Model.h5')

converter = tf.lite.TFLiteConverter.from_keras_model(model)

converter.target_spec.supported_ops = [tf.lite.OpsSet.TFLITE_BUILTINS, tf.lite.OpsSet.SELECT_TF_OPS]
converter._experimental_lower_tensor_list_ops = False

tflite_model = converter.convert()

output_directory = r'C:\Users\BAPPENAS\OneDrive\Documents\Bangkit\coba'

os.makedirs(output_directory, exist_ok=True)


output_file_path = os.path.join(output_directory, 'model.tflite')
with open(output_file_path, 'wb') as f:
    f.write(tflite_model)

print(f'TFLite model saved to: {output_file_path}')