R code for statistical time series analysis for the class STATGR5263 at Columbia University in Fall 21. Will reuse in future.
Mostly about transforming time series data to be (weakly) stationary (time invariant first and second central moments). Data is made stationary through
- Non-parametric smoothing:
- Kernel smoothing
- Moving Average smoothing
- Parametric smoothing:
- 2x Fitting a map to the model the mean through OLS and subtracting it from the data to handle the residuals like a weakly stationary time series.