This repository contains a comprehensive R project focused on time series analysis and forecasting using real-world economic and social datasets. The project demonstrates key techniques in time series decomposition, exponential smoothing, ARIMA modeling, dynamic regression, and hierarchical/grouped forecasting.

It is designed for learning, research, and as a reference for applying time series methods to different types of datasets.

• Time Series Decomposition

  • STL decomposition (including robust and seasonal variants)
  • Classical decomposition
  • X-11 and SEATS decomposition
  • Trend-cycle and seasonal analysis

• Exponential Smoothing

  • Simple Exponential Smoothing (SES)
  • Holt’s linear and damped trend methods
  • Holt-Winters additive and multiplicative seasonal methods
  • ETS model selection and forecasting

• ARIMA Modeling

  • Stationarity testing and differencing
  • Non-seasonal and seasonal ARIMA
  • Dynamic regression with lagged predictors

• Hierarchical and Grouped Time Series

  • Hierarchical Time Series (HTS)
  • Grouped Time Series (GTS)
  • Bottom-up and optimal reconciliation forecasting

• Visualization

  • Trend, seasonal, and remainder plots
  • Forecast plots with prediction intervals
  • Seasonal subseries and autocorrelation analysis

• R (>= 4.0)
• Recommended R packages:

  install.packages(c("forecast", "ggplot2", "tidyverse", "seasonal", "hts", "gridExtra", "urca"))