Learn Forecasting with Decomposition Insights

Swipe to show menu

Understanding how to use the trend and seasonality components extracted from time series decomposition can help you build simple yet surprisingly effective forecasting models. When you decompose a time series, you separate it into its underlying trend, repeating seasonal effects, and random residuals. By leveraging the trend and seasonality, you can construct a naive forecast: extend the trend into the future and add back the seasonal pattern, assuming both will continue as observed. This approach is especially useful for short-term forecasting where patterns are stable and changes are gradual.


              123456789101112131415161718192021222324252627282930313233343536373839404142434445
            
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from statsmodels.tsa.seasonal import seasonal_decompose

# Generate example time series data
np.random.seed(42)
periods = 48
time = np.arange(periods)
trend = 0.5 * time
seasonal = 10 * np.sin(2 * np.pi * time / 12)
noise = np.random.normal(scale=2, size=periods)
series = trend + seasonal + noise

# Create DataFrame
df = pd.DataFrame({'value': series})

# Decompose the time series
result = seasonal_decompose(df['value'], model='additive', period=12)
trend_component = result.trend
seasonal_component = result.seasonal

# Prepare for forecasting: extrapolate trend, repeat seasonality
forecast_horizon = 12
last_trend = trend_component.dropna().iloc[-1]
trend_slope = (trend_component.dropna().iloc[-1] - trend_component.dropna().iloc[0]) / (len(trend_component.dropna()) - 1)
future_trend = [last_trend + trend_slope * (i+1) for i in range(forecast_horizon)]

# Repeat last observed seasonality
seasonal_pattern = seasonal_component[-12:]
future_seasonality = list(seasonal_pattern.values)

# Naive forecast: trend + seasonality
future_forecast = np.array(future_trend) + np.array(future_seasonality)

# Plot
plt.figure(figsize=(10, 6))
plt.plot(df.index, df['value'], label='Original Series')
plt.plot(df.index, trend_component, label='Trend', linestyle='--')
plt.plot(np.arange(periods, periods + forecast_horizon), future_forecast, label='Naive Forecast', marker='o')
plt.legend()
plt.title('Naive Forecast Using Decomposed Trend and Seasonality')
plt.xlabel('Time')
plt.ylabel('Value')
plt.show()

While decomposition-based forecasting provides a quick and interpretable way to predict future values, it has important limitations. This approach assumes that both the trend and seasonal patterns will continue unchanged into the future, which is often not the case for real-world data prone to structural breaks or evolving behaviors. It is also sensitive to outliers and does not account for unpredictable events or changes in variance. However, this method is practical for establishing baseline forecasts, validating more complex models, or when you need transparent logic for short-term planning. Always compare its results with other methods and remember that decomposition-based forecasts are best for stable, well-understood time series.

Everything was clear?

Thanks for your feedback!

Section 1. Chapter 10

Ask AI

Ask anything or try one of the suggested questions to begin our chat

Section 1. Chapter 10