assignment05-week5

Olli Lehtonen

2026-03-27

Assignment 05: Geographically weighted regression

Use Statistics Finland’s Key Figures of Municipalities dataset in this exercise. Download the entire dataset with all variables for a single statistical year for all Finnish municipalities from the Statistics Finland website:

https://pxdata.stat.fi/PxWeb/pxweb/en/Kuntien_avainluvut/

When downloading the data, select the format “Comma delimited with heading”, so that the data are downloaded as a CSV table using a comma as the separator.

Join this dataset in R with the municipality dataset downloaded from the geofi package. You can save the key figures as a CSV table in your own directory and then import the data into R. You can import the CSV table into R, for example, using the read.csv function:

data_csv <- read.csv("data/mydata.csv")

The municipality dataset can be downloaded in R as follows:

municipalities25 <- geofi::get_municipalities(year = 2025)

After this, you can join the downloaded table to the municipality dataset (remember change variables names to correspond your data):

map <- left_join(municipalities25, data_csv, by = c("kunta" = "region")) # why we use left_join?

Use the Key Figures of Municipalities dataset in the tasks.

Task 1: Linear model

Create a linear regression model using the lm function where you explain the municipal unemployment rate using:

degree of urbanisation (%), workplace self-sufficiency, and share of persons with tertiary-level education (persons aged 15 or over, %).

Include the code you used in your answer and interpret the results of the model.

Task 2: Spatial autocorrelation

Continue from the previous task and test the residual spatial autocorrelation of the global linear model. How do you interpret the result?

Task 3: Geographically weighted regression

Fit a GWR model corresponding to the model in Task 1. Visualise the locally estimated regression coefficients on a map and provide an interpretation of the results.