Scale Industrial Forecasting with GluonTS and scikit-learn Ensemble Methods

                      
                     
from typing import List, Dict, Any
import os
import pandas as pd
import numpy as np
from gluonts.dataset.common import ListDataset
from gluonts.model.simple import SimpleFeedForward
from gluonts.evaluation import Evaluator
from sklearn.ensemble import VotingRegressor
from sklearn.linear_model import LinearRegression
from sklearn.tree import DecisionTreeRegressor
from sklearn.metrics import mean_squared_error

# Configuration
DATA_PATH = os.getenv('DATA_PATH', 'data.csv')
FORECAST_HORIZON = 30

# Load data
try:
    df = pd.read_csv(DATA_PATH)
    print("Data loaded successfully.")
except Exception as e:
    raise ValueError(f"Error loading data: {str(e)}")

# Prepare training dataset
training_data = ListDataset(
    [{'start': df['date'][0], 'target': df['value'][:len(df)//2]}],
    freq='D'
)

# Initialize GluonTS model
gluonts_model = SimpleFeedForward(num_hidden_dimensions=[10], prediction_length=FORECAST_HORIZON)

# Fit GluonTS model
try:
    gluonts_model.train(training_data)
    print("GluonTS model trained successfully.")
except Exception as e:
    raise RuntimeError(f"Error training GluonTS model: {str(e)}")

# Prepare scikit-learn ensemble model
model1 = LinearRegression()
model2 = DecisionTreeRegressor()
ensemble_model = VotingRegressor(estimators=[('lr', model1), ('dt', model2)])

# Fit the ensemble model
X = np.arange(len(df)//2).reshape(-1, 1)
Y = df['value'][:len(df)//2]
try:
    ensemble_model.fit(X, Y)
    print("Ensemble model trained successfully.")
except Exception as e:
    raise RuntimeError(f"Error training ensemble model: {str(e)}")

# Forecasting
try:
    predictions_gluonts = gluonts_model.predict(training_data)
    predictions_ensemble = ensemble_model.predict(np.arange(len(df)//2, len(df)//2 + FORECAST_HORIZON).reshape(-1, 1))
except Exception as e:
    raise RuntimeError(f"Error during forecasting: {str(e)}")

# Evaluate predictions
mse_gluonts = mean_squared_error(df['value'][len(df)//2:], predictions_gluonts)
mse_ensemble = mean_squared_error(df['value'][len(df)//2:], predictions_ensemble)
print(f"MSE GluonTS: {mse_gluonts}")
print(f"MSE Ensemble: {mse_ensemble}")

if __name__ == '__main__':
    print("Forecasting process completed.")