Task 1.

The first task is to calculate correlations between the temperature trace and the geographical parameters (and their statistical significance, of course!)

Watch BOTH videos before you do the exercise!

The hard way:

The easy way:

The files mentioned in the videos:

The Report Template: urban_climate_results_template_VT20.docx
The table of p-values for R and N: p-value-for-Pearson.xlsx

... and the equation used in Excel to calculate the p-value was:

=T.DIST.2T(ABS(R_cell)/SQRT((1 - R_cell*R_cell)/(N_cell-2)), (N_cell-2))

eller på svenska (och här gissa jag lite)
=T.FÖRD.2T(ABS(R_cell)/ROT((1 - R_cell*R_cell)/(N_cell-2)); (N_cell-2))

where you replace "R_cell" with the cell containing the R value (calculated with CORREL/KORREL, and N_cell with the cell containing the number of observations (calculated with COUNT/ANTAL). Note that in Swedish Excel, you need ; to separate input values to functions, whereas in English you need , !!

Task 2.

In this final part of the exercise, you make an urban climate map using multiple-regression.

Feel free to start a discussion thread where you share links to the best web-sites/video explanations for multiple-regression that you can find! Especially anything that you find in Swedish that is good that you think others would like.

We get the coefficients we need to make the urban temperature map:

...and then we use the coefficients in Qgis to make a map of the urban temperature!

Files that you might need:

Also, you can make the raster calculator step much faster by specifying the extent xmin, xmax, ymin, ymax.
Choose appropriate values for a box surrounding the measurements (eg by using the coordinates for the cursor position at the bottom of the QGIS window)