OBSOLETE - Materials moved to the StateSpaceRecipes repository
State-space models are useful for modeling time series data, and R contains several excellent packages for creating the models. Unfortunately, using these packages involves many little details, and I can never remember them. So I put together some recipes to let me quickly build state-space models.
I describe the recipes on my website.
This repository contains R code for the recipes. These are the main recipes, most likely to be useful for day-to-day work:
- localLevel.R - Local level model
- localLinearTrend.R - Local linear trend model
- regressionFixedCoeffs.R - Regression model with fixed coefficients
- regressionVaryingCoeffs.R - Regression model with time-varying coefficients
These recipes may come in handy in special circumstances:
- randomWalk.R - Random walk model, StructTS implementation
- randomWalk-dlm.R - Random walk model, dlm implemenation
- localLevel-dlm.R - Local level model, dlm implemenatation
- localLinearTrend-dlm.R - Local linear trend model, dlm implementation