Spaces:
Runtime error
Runtime error
File size: 2,110 Bytes
4ba82ce |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 |
from sklearn.ensemble import RandomForestRegressor
import pandas as pd
import numpy as np
def create_lagged_features(data, n_lags=5):
"""
Prepares the dataset with lagged features necessary for Random Forest regression.
Parameters:
- data: Pandas Series of historical closing prices.
- n_lags: The number of lagged observations to create as features.
Returns:
- A tuple (X, y) where X is a DataFrame of lagged features and y is the original dataset shifted.
"""
df = pd.DataFrame(data)
for lag in range(1, n_lags + 1):
df[f'lag_{lag}'] = df[data.name].shift(lag)
df.dropna(inplace=True) # Drop rows with NaN values resulted from shifting
X = df.drop(columns=[data.name])
y = df[data.name]
return X, y
def random_forest_forecast(data, forecast_horizon, n_lags=5):
"""
Forecast future values using a Random Forest Regressor, with a dynamic forecast horizon.
Parameters:
- data: Pandas Series of historical closing prices.
- forecast_horizon: Integer specifying the number of days to forecast.
- n_lags: Number of past observations to use for forecasting.
Returns:
- Pandas Series containing the forecasted values with a datetime index.
"""
X, y = create_lagged_features(data, n_lags)
model = RandomForestRegressor(n_estimators=100, random_state=42)
model.fit(X, y)
# Prepare the data for forecasting
last_obs = data.tail(n_lags).values[::-1] # Reverse to get the correct order (most recent first)
forecasts = []
for _ in range(forecast_horizon):
# Reshape last_obs to match model input shape
model_input = np.array(last_obs).reshape(1, -1)
forecast = model.predict(model_input)[0]
forecasts.append(forecast)
# Update last_obs with the forecasted value
last_obs = np.roll(last_obs, -1)
last_obs[-1] = forecast
future_dates = pd.date_range(start=data.index[-1] + pd.Timedelta(days=1), periods=forecast_horizon)
forecast_series = pd.Series(forecasts, index=future_dates)
return forecast_series
|