Instructors:
Gregory Britten ([email protected])
Yara Mohajerani ([email protected])
Department of Earth System Science
University of California, Irvine
This one-day workshop will introduce you to concepts from statistics that are most important for statistically analyzing spatial-temporal data. Motivated by datasets from the envrionmental sciences, we focus our study on two concepts: statistical autocorrelation and periodicity. The meaning of these terms will become clear as we work through the notebooks.
The workshop is divided in three basic parts: theory, software, and hands-on data analysis. The theory will build upon basic statistics to extend concepts to spatial-temporal properties. For software we'll use a few key R packages, along with Python functionality to access R packages from within Python.
The schedule of the course is as follows:
- Intall instructions are available from the link above 'Intro : Requirements.ipynb'
- PLEASE COME TO THE WORKSHOP ROOM EARLY (8:30) IF ANY PART OF THE INSTALL INSTRUCTIONS FAILS
- Introduction to spatial-temporal statistics
- Hands-on simulation to understand statistical concepts
- Applied time series analysis
- Trend analysis, harmonic regression (for seasonality and other peiodicity), regression with covariates
- Statistical model selection for time series
- Extension of time series ideas to spatial analysis
- Variogram functions
- Spatial interpolation (Krigeing)
- Spatial regression
- Model selection for spatial models
These folders contain the Jupyter notebooks. The notebooks contain notes to complement the slide presentation, along with executable code to implement the analyses we discuss. This is also the place where you will extend the analyses in hands-on exercises. The R and Python folders are effectively mirror copies of one another for the respective languages.
- El Nino time series
- Orange County air quality time series
- Oregon climate station data