What problems can a student solve when completing
the course?
- Take a set of almost-linear data and use a spreadsheet to
find the regression line and then do all the same things they would be able
to do if the relationship was exactly linear, PLUS use a residual plot to
explain why a linear model is appropriate.
- Take a set of data which is clearly non-linear and experiment with various
non-linear models to find one that fits reasonably well.
- Produce and interpret a residual plot for that.
- Use that non-linear model and the graph/ spreadsheet to make predictions
of y from x and also to find what x-values give a particular
y-value.
- Understand that the variability around the line that is shown in the residual
plot is “noise” rather than “pattern” and that this noise can often be thought
of as measurement noise.