df = pd.read_csv('dataset.csv', sep=';') df_melted = pd.melt(df, id_vars=['komoditas'], var_name='tanggal', value_name='harga') df_melted.head() df_beras = df_melted[df_melted['komoditas'] == 'Beras'].drop(columns=['komoditas']) df_gula = df_melted[df_melted['komoditas'] == 'Gula Pasir'].drop(columns=['komoditas']) df_minyak = df_melted[df_melted['komoditas'] == 'Minyak Goreng'].drop(columns=['komoditas']) df_beras['tanggal'] = pd.to_datetime(df_beras['tanggal'], format='%d/%m/%Y').dt.strftime('%Y-%m-%d') df_gula['tanggal'] = pd.to_datetime(df_gula['tanggal'], format='%d/%m/%Y').dt.strftime('%Y-%m-%d') df_minyak['tanggal'] = pd.to_datetime(df_minyak['tanggal'], format='%d/%m/%Y').dt.strftime('%Y-%m-%d')

def windowed_dataset(series, window_size, batch_size, shuffle_buffer): series = tf.expand_dims(series, axis=-1) ds = tf.data.Dataset.from_tensor_slices(series) ds = ds.window(window_size + 1, shift=1, drop_remainder=True) ds = ds.flat_map(lambda w: w.batch(window_size + 1)) ds = ds.shuffle(shuffle_buffer) ds = ds.map(lambda w: (w[:-1], w[1:])) return ds.batch(batch_size).prefetch(1)

def model_fit(df): time_step = df['tanggal'].values prices = df['harga'].values series = np.array(prices) min = np.min(series) max = np.max(series) series -= min series /= max time = np.array(time_step) split_time = 3000 time_train = time[:split_time] x_train = series[:split_time] time_valid = time[split_time:] x_valid = series[split_time:] window_size = 30 batch_size = 32 shuffle_buffer_size = 1000 train_set = windowed_dataset(x_train, window_size=window_size, batch_size=batch_size, shuffle_buffer=shuffle_buffer_size) model = tf.keras.models.Sequential([ tf.keras.layers.Conv1D(filters=32, kernel_size=5, strides=1, padding="causal", activation="relu", input_shape=[None, 1]), tf.keras.layers.LSTM(64, return_sequences=True), tf.keras.layers.LSTM(64, return_sequences=True), tf.keras.layers.Dense(30, activation="relu"), tf.keras.layers.Dense(10, activation="relu"), tf.keras.layers.Dense(1) ]) model.compile(loss=tf.keras.losses.Huber(), optimizer=tf.keras.optimizers.SGD(learning_rate=1e-5, momentum=0.9), metrics=["mae"]) model.fit(train_set, epochs=100) return model

safrom datetime import datetime, timedelta def predict_future_price_on_date(model, df, target_date_str, window_size): # Konversi tanggal target ke datetime target_date = datetime.strptime(target_date_str, '%Y-%m-%d') # Cari tanggal terakhir dalam dataset last_date_str = df['tanggal'].iloc[-1] last_date = datetime.strptime(last_date_str, '%Y-%m-%d') # Hitung selisih hari antara tanggal terakhir dan tanggal target delta_days = (target_date - last_date).days if delta_days <= 0: raise ValueError("Target date must be in the future.") # Normalisasi harga min_price = df['harga'].min() max_price = df['harga'].max() normalized_prices = (df['harga'] - min_price) / (max_price - min_price) # Prediksi harga untuk setiap hari hingga tanggal target current_date = last_date for _ in range(delta_days): # Gunakan data harga terakhir sebanyak window_size last_data_points = normalized_prices[-window_size:].values.reshape((1, window_size, 1)) # Prediksi harga untuk hari berikutnya predicted_normalized_price = model.predict(last_data_points) # Tambahkan prediksi ke daftar harga yang dinormalisasi normalized_prices = normalized_prices.append(pd.Series(predicted_normalized_price.ravel()[0])) # Perbarui tanggal saat ini current_date += timedelta(days=1) # Denormalisasi harga prediksi terakhir predicted_price = normalized_prices.iloc[-1] * (max_price - min_price) + min_price return predicted_price

window_size = 30

target_date = target_date.strftime('%Y-%m-%d')