File size: 13,598 Bytes
26a8ea5 |
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 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 |
"""
Economic Forecasting Module
Advanced time series forecasting for economic indicators using ARIMA/ETS models
"""
import logging
import warnings
from datetime import datetime, timedelta
from typing import Dict, List, Optional, Tuple, Union
import numpy as np
import pandas as pd
from scipy import stats
from sklearn.metrics import mean_absolute_error, mean_squared_error
from statsmodels.tsa.arima.model import ARIMA
from statsmodels.tsa.holtwinters import ExponentialSmoothing
from statsmodels.tsa.seasonal import seasonal_decompose
from statsmodels.tsa.stattools import adfuller
logger = logging.getLogger(__name__)
class EconomicForecaster:
"""
Advanced economic forecasting using ARIMA and ETS models
with comprehensive backtesting and performance evaluation
"""
def __init__(self, data: pd.DataFrame):
"""
Initialize forecaster with economic data
Args:
data: DataFrame with economic indicators (GDPC1, INDPRO, RSAFS, etc.)
"""
self.data = data.copy()
self.forecasts = {}
self.backtest_results = {}
self.model_performance = {}
def prepare_data(self, target_series: str, frequency: str = 'Q') -> pd.Series:
"""
Prepare time series data for forecasting
Args:
target_series: Series name to forecast
frequency: Data frequency ('Q' for quarterly, 'M' for monthly)
Returns:
Prepared time series
"""
if target_series not in self.data.columns:
raise ValueError(f"Series {target_series} not found in data")
series = self.data[target_series].dropna()
# Resample to desired frequency
if frequency == 'Q':
series = series.resample('Q').mean()
elif frequency == 'M':
series = series.resample('M').mean()
# Calculate growth rates for economic indicators
if target_series in ['GDPC1', 'INDPRO', 'RSAFS']:
series = series.pct_change().dropna()
return series
def check_stationarity(self, series: pd.Series) -> Dict:
"""
Perform Augmented Dickey-Fuller test for stationarity
Args:
series: Time series to test
Returns:
Dictionary with test results
"""
result = adfuller(series.dropna())
return {
'adf_statistic': result[0],
'p_value': result[1],
'critical_values': result[4],
'is_stationary': result[1] < 0.05
}
def decompose_series(self, series: pd.Series, period: int = 4) -> Dict:
"""
Decompose time series into trend, seasonal, and residual components
Args:
series: Time series to decompose
period: Seasonal period (4 for quarterly, 12 for monthly)
Returns:
Dictionary with decomposition components
"""
decomposition = seasonal_decompose(series.dropna(), period=period, extrapolate_trend='freq')
return {
'trend': decomposition.trend,
'seasonal': decomposition.seasonal,
'residual': decomposition.resid,
'observed': decomposition.observed
}
def fit_arima_model(self, series: pd.Series, order: Tuple[int, int, int] = None) -> ARIMA:
"""
Fit ARIMA model to time series
Args:
series: Time series data
order: ARIMA order (p, d, q). If None, auto-detect
Returns:
Fitted ARIMA model
"""
if order is None:
# Auto-detect order using AIC minimization
best_aic = np.inf
best_order = (1, 1, 1)
for p in range(0, 3):
for d in range(0, 2):
for q in range(0, 3):
try:
model = ARIMA(series, order=(p, d, q))
fitted_model = model.fit()
if fitted_model.aic < best_aic:
best_aic = fitted_model.aic
best_order = (p, d, q)
except:
continue
order = best_order
logger.info(f"Auto-detected ARIMA order: {order}")
model = ARIMA(series, order=order)
fitted_model = model.fit()
return fitted_model
def fit_ets_model(self, series: pd.Series, seasonal_periods: int = 4) -> ExponentialSmoothing:
"""
Fit ETS (Exponential Smoothing) model to time series
Args:
series: Time series data
seasonal_periods: Number of seasonal periods
Returns:
Fitted ETS model
"""
model = ExponentialSmoothing(
series,
seasonal_periods=seasonal_periods,
trend='add',
seasonal='add'
)
fitted_model = model.fit()
return fitted_model
def forecast_series(self, series: pd.Series, model_type: str = 'auto',
forecast_periods: int = 4) -> Dict:
"""
Forecast time series using specified model
Args:
series: Time series to forecast
model_type: 'arima', 'ets', or 'auto'
forecast_periods: Number of periods to forecast
Returns:
Dictionary with forecast results
"""
if model_type == 'auto':
# Try both models and select the one with better AIC
try:
arima_model = self.fit_arima_model(series)
arima_aic = arima_model.aic
except:
arima_aic = np.inf
try:
ets_model = self.fit_ets_model(series)
ets_aic = ets_model.aic
except:
ets_aic = np.inf
if arima_aic < ets_aic:
model_type = 'arima'
model = arima_model
else:
model_type = 'ets'
model = ets_model
elif model_type == 'arima':
model = self.fit_arima_model(series)
elif model_type == 'ets':
model = self.fit_ets_model(series)
else:
raise ValueError("model_type must be 'arima', 'ets', or 'auto'")
# Generate forecast
forecast = model.forecast(steps=forecast_periods)
# Calculate confidence intervals
if model_type == 'arima':
forecast_ci = model.get_forecast(steps=forecast_periods).conf_int()
else:
# For ETS, use simple confidence intervals
forecast_std = series.std()
forecast_ci = pd.DataFrame({
'lower': forecast - 1.96 * forecast_std,
'upper': forecast + 1.96 * forecast_std
})
return {
'model': model,
'model_type': model_type,
'forecast': forecast,
'confidence_intervals': forecast_ci,
'aic': model.aic if hasattr(model, 'aic') else None
}
def backtest_forecast(self, series: pd.Series, model_type: str = 'auto',
train_size: float = 0.8, test_periods: int = 8) -> Dict:
"""
Perform backtesting of forecasting models
Args:
series: Time series to backtest
model_type: Model type to use
train_size: Proportion of data for training
test_periods: Number of periods to test
Returns:
Dictionary with backtest results
"""
n = len(series)
train_end = int(n * train_size)
actual_values = []
predicted_values = []
errors = []
for i in range(test_periods):
if train_end + i >= n:
break
# Use expanding window
train_data = series.iloc[:train_end + i]
test_value = series.iloc[train_end + i]
try:
forecast_result = self.forecast_series(train_data, model_type, 1)
prediction = forecast_result['forecast'].iloc[0]
actual_values.append(test_value)
predicted_values.append(prediction)
errors.append(test_value - prediction)
except Exception as e:
logger.warning(f"Forecast failed at step {i}: {e}")
continue
if not actual_values:
return {'error': 'No successful forecasts generated'}
# Calculate performance metrics
mae = mean_absolute_error(actual_values, predicted_values)
mse = mean_squared_error(actual_values, predicted_values)
rmse = np.sqrt(mse)
mape = np.mean(np.abs(np.array(actual_values) - np.array(predicted_values)) / np.abs(actual_values)) * 100
return {
'actual_values': actual_values,
'predicted_values': predicted_values,
'errors': errors,
'mae': mae,
'mse': mse,
'rmse': rmse,
'mape': mape,
'test_periods': len(actual_values)
}
def forecast_economic_indicators(self, indicators: List[str] = None) -> Dict:
"""
Forecast multiple economic indicators
Args:
indicators: List of indicators to forecast. If None, use default set
Returns:
Dictionary with forecasts for all indicators
"""
if indicators is None:
indicators = ['GDPC1', 'INDPRO', 'RSAFS']
results = {}
for indicator in indicators:
try:
# Prepare data
series = self.prepare_data(indicator)
# Check stationarity
stationarity = self.check_stationarity(series)
# Decompose series
decomposition = self.decompose_series(series)
# Generate forecast
forecast_result = self.forecast_series(series)
# Perform backtest
backtest_result = self.backtest_forecast(series)
results[indicator] = {
'stationarity': stationarity,
'decomposition': decomposition,
'forecast': forecast_result,
'backtest': backtest_result,
'series': series
}
logger.info(f"Successfully forecasted {indicator}")
except Exception as e:
logger.error(f"Failed to forecast {indicator}: {e}")
results[indicator] = {'error': str(e)}
return results
def generate_forecast_report(self, forecasts: Dict) -> str:
"""
Generate comprehensive forecast report
Args:
forecasts: Dictionary with forecast results
Returns:
Formatted report string
"""
report = "ECONOMIC FORECASTING REPORT\n"
report += "=" * 50 + "\n\n"
for indicator, result in forecasts.items():
if 'error' in result:
report += f"{indicator}: ERROR - {result['error']}\n\n"
continue
report += f"INDICATOR: {indicator}\n"
report += "-" * 30 + "\n"
# Stationarity results
stationarity = result['stationarity']
report += f"Stationarity Test (ADF):\n"
report += f" ADF Statistic: {stationarity['adf_statistic']:.4f}\n"
report += f" P-value: {stationarity['p_value']:.4f}\n"
report += f" Is Stationary: {stationarity['is_stationary']}\n\n"
# Model information
forecast = result['forecast']
report += f"Model: {forecast['model_type'].upper()}\n"
if forecast['aic']:
report += f"AIC: {forecast['aic']:.4f}\n"
report += f"Forecast Periods: {len(forecast['forecast'])}\n\n"
# Backtest results
backtest = result['backtest']
if 'error' not in backtest:
report += f"Backtest Performance:\n"
report += f" MAE: {backtest['mae']:.4f}\n"
report += f" RMSE: {backtest['rmse']:.4f}\n"
report += f" MAPE: {backtest['mape']:.2f}%\n"
report += f" Test Periods: {backtest['test_periods']}\n\n"
# Forecast values
report += f"Forecast Values:\n"
for i, value in enumerate(forecast['forecast']):
ci = forecast['confidence_intervals']
lower = ci.iloc[i]['lower'] if 'lower' in ci.columns else 'N/A'
upper = ci.iloc[i]['upper'] if 'upper' in ci.columns else 'N/A'
report += f" Period {i+1}: {value:.4f} [{lower:.4f}, {upper:.4f}]\n"
report += "\n" + "=" * 50 + "\n\n"
return report |