Update app.py

app.py

@@ -1,9 +1,10 @@
 from fastapi import FastAPI
+from asgiref.wsgi import WsgiToAsgi
 from flask import Flask, render_template, request, jsonify
-from fastapi.middleware.wsgi import WSGIMiddleware
 import pickle
 import numpy as np
 import pandas as pd
+from fastapi.middleware.wsgi import WSGIMiddleware
 
 flask_app = Flask(__name__)
 
@@ -14,64 +15,71 @@ with open('model.pkl', 'rb') as model_file:
 with open('scaler.pkl', 'rb') as scaler_file:
     scaler = pickle.load(scaler_file)
 
-# Fungsi untuk membentuk windowed data
-def create_windowed_data(sell_features, buy_features):
-    # Membuat DataFrame dengan lag 7 dari fitur sell dan buy
-    windowed_data = pd.DataFrame({
-        'sell-1': [sell_features[6]],
-        'sell-2': [sell_features[5]],
-        'sell-3': [sell_features[4]],
-        'sell-4': [sell_features[3]],
-        'sell-5': [sell_features[2]],
-        'sell-6': [sell_features[1]],
-        'sell-7': [sell_features[0]],
-        'buy-1': [buy_features[6]],
-        'buy-2': [buy_features[5]],
-        'buy-3': [buy_features[4]],
-        'buy-4': [buy_features[3]],
-        'buy-5': [buy_features[2]],
-        'buy-6': [buy_features[1]],
-        'buy-7': [buy_features[0]],
-    })
-    return windowed_data
-
 # Fungsi prediksi untuk 7 hari ke depan
-def predict_7_days(windowed_data, linear_model, scaler):
-    # Normalisasi data fitur
-    normalized_data = scaler.transform(windowed_data.values)
-
-    # Ambil baris terakhir sebagai input awal
-    last_row = normalized_data[-1].reshape(1, -1)
+def predict_7_days(sell_features, buy_features, linear_model, scaler):
+    # Data dari frontend
+    # Ambil nilai input (pastikan semuanya adalah angka)
+    sell_1 = float(sell_features[6])
+    sell_2 = float(sell_features[5])
+    sell_3 = float(sell_features[4])
+    sell_4 = float(sell_features[3])
+    sell_5 = float(sell_features[2])
+    sell_6 = float(sell_features[1])
+    sell_7 = float(sell_features[0])
+
+    buy_1 = float(buy_features[6])
+    buy_2 = float(buy_features[5])
+    buy_3 = float(buy_features[4])
+    buy_4 = float(buy_features[3])
+    buy_5 = float(buy_features[2])
+    buy_6 = float(buy_features[1])
+    buy_7 = float(buy_features[0])
+
+    # Menyusun data jual dan beli sebagai pasangan (sell, buy)
+    data = np.array([
+        [sell_1, buy_1],
+        [sell_2, buy_2],
+        [sell_3, buy_3],
+        [sell_4, buy_4],
+        [sell_5, buy_5],
+        [sell_6, buy_6],
+        [sell_7, buy_7]
+    ])
+    
+    # Normalisasi data menggunakan scaler (menormalkan pasangan jual dan beli)
+    normalized_data = scaler.transform(data)
+    
+    # Buat DataFrame dari nilai yang telah dinormalisasi
+    # Kolom disesuaikan untuk 7 fitur jual dan 7 fitur beli
+    last_row = pd.DataFrame(
+        normalized_data.flatten().reshape(1, -1), 
+        columns=['sell-1', 'sell-2', 'sell-3', 'sell-4', 'sell-5', 'sell-6', 'sell-7', 'buy-1', 'buy-2', 'buy-3', 'buy-4', 'buy-5', 'buy-6', 'buy-7']
+    )
 
-    # Inisialisasi untuk menyimpan prediksi
     predictions = []
 
     # Iterasi untuk 7 hari ke depan
     for _ in range(7):
-        # Prediksi nilai normalisasi untuk hari berikutnya
+        # Prediksi nilai untuk hari berikutnya
         predicted_value_normalized = linear_model.predict(last_row)
-
-        # Inverse transform hasil prediksi untuk mendapatkan nilai asli
-        predicted_value = scaler.inverse_transform(
-            np.hstack([last_row[0, 2:], predicted_value_normalized[0]]).reshape(1, -1)
-        )[:, -2:]  # Ambil hanya kolom 'sell' dan 'buy'
-
-        # Simpan hasil prediksi
+        predicted_value = scaler.inverse_transform(predicted_value_normalized.reshape(-1, 2))
+        
+        # Simpan prediksi
         predictions.append(predicted_value[0])
+        
+        # Update input untuk iterasi berikutnya
+        new_row_normalized = np.hstack([last_row.iloc[0, 2:], predicted_value_normalized[0]])
+        last_row = pd.DataFrame(new_row_normalized.reshape(1, -1), columns=last_row.columns)
 
-        # Perbarui `last_row` untuk prediksi berikutnya
-        new_row_normalized = np.hstack([last_row[0, 2:], predicted_value_normalized[0]])
-        last_row = new_row_normalized.reshape(1, -1)
-
-    # Buat DataFrame untuk hasil prediksi
+    # Transformasikan prediksi menjadi DataFrame untuk visualisasi
     predictions_df = pd.DataFrame(
-        predictions,
-        columns=['sell', 'buy'],
+        predictions, 
+        columns=['sell', 'buy'], 
         index=pd.date_range(start=pd.Timestamp.today(), periods=7)
     )
 
     # Harga terakhir
-    last_price = scaler.inverse_transform(normalized_data[-1].reshape(1, -1))[:, -2:]
+    last_price = scaler.inverse_transform(normalized_data[-1].reshape(-1, 2))
 
     # Hitung persentase perubahan harian
     predictions_df['sell_change'] = predictions_df['sell'].pct_change().fillna(0) * 100
@@ -96,11 +104,8 @@ def predict():
     sell_features = data['sell_features']
     buy_features = data['buy_features']
 
-    # Bentuk data menjadi windowed_data
-    windowed_data = create_windowed_data(sell_features, buy_features)
-
     # Prediksi harga untuk 7 hari ke depan
-    predictions_df, total_sell_change, total_buy_change, last_price = predict_7_days(windowed_data, linear_model, scaler)
+    predictions_df, total_sell_change, total_buy_change, last_price = predict_7_days(sell_features, buy_features, linear_model, scaler)
 
     # Membuat hasil prediksi untuk respons JSON
     predictions_result = []