Key Forecasting Methods to Know for Intro to Time Series

Forecasting methods help us predict future values in time series data by analyzing past observations. Techniques like Moving Average, ARIMA, and Exponential Smoothing allow us to identify trends, seasonality, and patterns, making our forecasts more accurate and reliable.

1. Moving Average (MA)

   • Averages a set number of past observations to smooth out short-term fluctuations.
   • Useful for identifying trends over time by reducing noise in the data.
   • Can be simple (unweighted) or weighted, where more recent observations have greater influence.

2. Exponential Smoothing

   • Applies decreasing weights to past observations, giving more importance to recent data.
   • Suitable for data with no clear trend or seasonal pattern.
   • Variants include simple, double, and triple exponential smoothing to account for trends and seasonality.

3. Autoregressive (AR) Models

   • Predict future values based on a linear combination of past values in the series.
   • The model order (p) indicates how many past observations are used.
   • Assumes that past values have a direct influence on future values.

4. Autoregressive Integrated Moving Average (ARIMA)

   • Combines autoregressive and moving average components with differencing to make the series stationary.
   • Requires identification of parameters (p, d, q) for effective modeling.
   • Suitable for non-seasonal time series data with trends.

5. Seasonal ARIMA (SARIMA)

   • Extends ARIMA by incorporating seasonal components into the model.
   • Parameters include seasonal order (P, D, Q) in addition to non-seasonal parameters.
   • Effective for time series data exhibiting seasonal patterns.

6. Trend Analysis

   • Involves identifying and analyzing long-term movements in data over time.
   • Can be visualized using line graphs or fitted trend lines.
   • Helps in understanding the direction and strength of trends for forecasting.

7. Decomposition Methods

   • Breaks down a time series into its constituent components: trend, seasonality, and residuals.
   • Useful for understanding underlying patterns and improving forecasting accuracy.
   • Can be additive or multiplicative, depending on the relationship between components.

8. Holt-Winters Method

   • An extension of exponential smoothing that accounts for both trends and seasonality.
   • Includes three smoothing parameters: level, trend, and seasonal.
   • Effective for forecasting data with seasonal patterns and trends.

9. Regression Analysis

   • Models the relationship between a dependent variable and one or more independent variables.
   • Can be used for time series forecasting by including time as an independent variable.
   • Assumes a linear relationship, but can be adapted for non-linear relationships.

10. Prophet (Facebook's forecasting tool)

    • Designed for forecasting time series data that may have missing values and outliers.
    • Utilizes an additive model with components for trend, seasonality, and holidays.
    • User-friendly interface allows for easy adjustments and customization of forecasts.