90% complete

test_hybrid.py

+import numpy as np
+import pandas as pd
+from keras.models import load_model
+from statsmodels.tsa.arima.model import ARIMAResults
+from datetime import date, timedelta
+import warnings
+
+warnings.filterwarnings("ignore")
+
+df = pd.read_csv('out_new.csv')
+df.Month = pd.to_datetime(df.Month)
+df = df.set_index("Month")
+
+train_data = df[:len(df) - 12]
+test_data = df[len(df) - 12:]
+
+next_12_months = []
+rain_zero_data = np.zeros(12, int)
+
+first_day_of_this_month = date.today().replace(day=1)
+next_12_months.append(first_day_of_this_month.strftime("%Y-%m-%d"))
+
+for i in range(1, 12):
+    if i == 1:
+        temp = first_day_of_this_month
+    start_day_of_next_month = (temp + timedelta(days=32)).replace(day=1)
+    temp = start_day_of_next_month
+
+    # For printing results
+    next_12_months.append(temp.strftime("%Y-%m-%d"))
+
+print(next_12_months)
+print(rain_zero_data)
+
+x = pd.DataFrame(list(zip(next_12_months, rain_zero_data)), columns=['Month', 'Monthly rain'])
+x.Month = pd.to_datetime(x.Month)
+x = x.set_index("Month")
+
+print('x type', type(x))
+print('test type', type(test_data))
+
+print('x info', x.info())
+print('test data info', test_data.info())
+
+arima_pred_result = []
+lstm_pred_result = []
+fbp_pred_result = []
+
+print('-------------------arima----------------------')
+
+arima_result = ARIMAResults.load('hybrid_trained_models/arima_model.pkl')
+
+arima_pred = arima_result.predict(start=len(df), end=len(df) + 11, typ="levels").rename("ARIMA Predictions")
+print(arima_pred)
+
+for x in range(len(arima_pred)):
+    arima_pred_result.append(round(arima_pred[x], 2))
+
+print('-------------------lstm----------------------')
+from sklearn.preprocessing import MinMaxScaler
+
+scaler = MinMaxScaler()
+scaler.fit(train_data)
+scaled_train_data = scaler.transform(df)
+
+lstm_model = load_model('hybrid_trained_models/lstm_model.h5')
+n_input = 12
+n_features = 1
+
+lstm_predictions_scaled = list()
+
+batch = scaled_train_data[-n_input:]
+current_batch = batch.reshape((1, n_input, n_features))
+
+for i in range(len(test_data)):
+    lstm_pred = lstm_model.predict(current_batch)[0]
+    lstm_predictions_scaled.append(lstm_pred)
+    current_batch = np.append(current_batch[:, 1:, :], [[lstm_pred]], axis=1)
+
+lstm_predictions = scaler.inverse_transform(lstm_predictions_scaled)
+
+lstm_predictions_list = lstm_predictions.tolist()
+print('type', type(lstm_predictions_list))
+print(lstm_predictions_list)
+
+for x in range(len(lstm_predictions_list)):
+    lstm_pred_result.append(round(lstm_predictions_list[x][0], 2))
+
+print('-------------------fb----------------------')
+
+from prophet.serialize import model_to_json, model_from_json
+
+with open('hybrid_trained_models/fbp_model.json', 'r') as fin:
+    m = model_from_json(fin.read())  # Load model
+
+future = m.make_future_dataframe(periods=12, freq='MS')
+prophet_pred = m.predict(future)
+
+for i in prophet_pred[-12:]["yhat"]:
+    fbp_pred_result.append(round(i, 2))
+
+prophet_pred = pd.DataFrame({"Date": prophet_pred[-12:]['ds'], "Pred": prophet_pred[-12:]["yhat"]})
+prophet_pred = prophet_pred.set_index("Date")
+prophet_pred.index.freq = "MS"
+print(prophet_pred)
+
+print('-------------------hybrid result----------------------')
+
+print('arima result : ')
+print(arima_pred_result)
+print('lstm result : ')
+print(lstm_pred_result)
+print('fbp result : ')
+print(fbp_pred_result)
+
+hybrid_result_list = []
+
+for i in range(len(arima_pred)):
+    hybrid_result_list.append(round((arima_pred_result[i] + lstm_pred_result[i] + fbp_pred_result[i]) / 3, 2))
+
+print('Hybrid result : ')
+print(hybrid_result_list)