Course Title

R Programming for Spatial Statistics & Modelling: Beginner to Advanced

---

Course Objectives

• Learn to handle spatial vector and raster data in R.
• Perform Exploratory Spatial Data Analysis (ESDA) and visualize spatial patterns.
• Understand and apply geostatistical methods (Kriging, Variograms).
• Use Spatial Regression Models (SAR, CAR, SEM) for areal data.
• Analyze Point Patterns and perform spatial autocorrelation tests.
• Implement Machine Learning (ML) and Generalized Additive Models (GAM) for spatial predictions.
• Apply Spatial Cross-Validation for reliable model evaluation.
• Generate publication-quality static and interactive maps.

---

Course Duration

7 Days – Hands-on training with practical exercises each day

Private training fee 1 vs 1: 800 USD

---

Day-wise Syllabus

Day 1: Introduction & Spatial Data Handling in R

Topics Covered:

• Introduction to R for spatial analysis
• Spatial data types: Raster vs Vector
• Coordinate Reference Systems (CRS) and projections
• Reading/Writing spatial data (Shapefiles, GeoTIFFs, GeoPackages)
• Basic raster operations: crop, mask, zonal statistics

Hands-on Lab:

• Using sf and terra for vector/raster operations
• Clipping, masking, and extracting zonal statistics

Deliverables:

• PNG map of clipped raster
• CSV table of zonal statistics

---

Day 2: Exploratory Spatial Data Analysis (ESDA) & Mapping

Topics Covered:

• Spatial descriptive statistics
• Data classification methods: Quantile, Equal Interval, Jenks
• Choropleth mapping using tmap and ggplot2
• Interactive mapping using leaflet
• Basic exploratory plots: histograms, boxplots, density plots

Hands-on Lab:

• Thematic maps using tmap and classInt
• Interactive maps with leaflet

Deliverables:

• PDF report with EDA figures
• HTML interactive map

---

Day 3: Spatial Autocorrelation & Variogram Analysis

Topics Covered:

• Spatial autocorrelation concepts
• Global Moran’s I and Local Moran’s I (LISA)
• Semivariogram and Variogram modeling
• Spatial dependency analysis

Hands-on Lab:

• Compute Global and Local Moran’s I using spdep
• Create and fit variogram models using gstat

Deliverables:

• Variogram model plot
• Moran’s I results with p-values

---

Day 4: Geostatistical Modelling (Kriging)

Topics Covered:

• Ordinary Kriging (OK)
• Universal Kriging (UK) with trend models
• Kriging variance and uncertainty mapping
• Generating prediction rasters

Hands-on Lab:

• Perform OK and UK using gstat
• Create prediction and variance maps

Deliverables:

• GeoTIFF of kriging prediction
• RMSE of kriging model using validation points

---

Day 5: Areal Data Modelling (SAR, CAR, SEM)

Topics Covered:

• Spatial weights matrices (contiguity, distance)
• Spatial Lag Models (SAR)
• Spatial Error Models (SEM)
• Conditional & Simultaneous Autoregressive Models (CAR, SAR)
• Model diagnostics & residual analysis

Hands-on Lab:

• Fit SAR & SEM models using spatialreg
• Compare AIC, residual Moran’s I, LM tests

Deliverables:

• Model summary table with diagnostics
• Map of residuals and fitted values

---

Day 6: Point Pattern Analysis & Machine Learning

Topics Covered:

• Point Pattern Analysis: Ripley’s K, density estimation
• Spatial Cross-Validation using blockCV
• Generalized Additive Models (GAM) with spatial smooths
• Random Forest & ML-based spatial prediction

Hands-on Lab:

• Kernel density maps using spatstat
• 5-fold spatial CV for Random Forest and GAM
• Prediction mapping using ranger and mgcv

Deliverables:

• Accuracy table (RMSE, MAE) for ML models
• GeoTIFF of best model prediction

---

Day 7: Model Comparison, Visualization & Capstone Project

Topics Covered:

• Comparing Kriging, GAM, RF, and SAR models
• Partial Dependence Plots & Variable Importance
• Interactive visualization of prediction maps
• Capstone project workflow: Data → Modeling → Validation → Prediction

Hands-on Lab:

• Build final model on full dataset
• Export results as GeoTIFF, PNG maps, and interactive Leaflet maps

Capstone Project:

• Choose one real-world dataset (e.g., Soil Moisture, LST, NDVI)
• Apply full workflow: ESDA → Modeling → Cross-validation → Prediction

Deliverables:

• Final project report (PDF) with methods, results, maps
• All scripts and GeoTIFF outputs

---

Software & Tools

• R version: Latest
• R Packages: sf, terra, tmap, ggplot2, classInt, spdep, spatialreg, gstat, automap, spatstat.explore, spatstat.model, blockCV, mgcv, ranger, tidymodels, dplyr, purrr

---

Evaluation

• Daily practical assignments: 50%
• Final Capstone Project: 50%

---

Expected Outcomes

By the end of this course, participants will be able to:

• Handle and analyze spatial datasets in R
• Perform geostatistical and spatial regression modeling
• Implement machine learning with spatial cross-validation
• Create high-quality static and interactive spatial maps
• Develop reproducible workflows for spatial data science projects