G'day,
I am a tutor in Further Mathematics, but have a question for fellow experts and students.
The picture attached is of a question my student attempted. The content is on chapter 3 of cambridge, so is on least squares regression. A few issues:
- The original data has a massive outlier, so shouldn't even be subject to least squares regression per the assumptions
- The picture is of the residual plot, which to me has no real observable pattern given the outlier, however perhaps the definition of "pattern" is not so clear cut? I imagine the pattern being non-linear or curved or some such, so how do we distinguish between pattern and non pattern?
Thanks for the help. Hopefully someone more zen than I can help me understand why we'd approach this question despite the outlier being present!
Cheers
hi! this is quite a late reply haha but i think that's what they're actually asking for in terms of why a regression analysis is not appropriate for the data. like you said there's a very clear outlier, and also there seems to be a decreasing pattern and somewhat linearity in the rest of the residual plot. ofc a pattern in the residual plot indicates the data is non-linear and hence further supports why linear regression is not suitable for the data. i feel like you just have to go with the information they've provided and then use your observations to answer part g