Question

When calculating the LSRL, should you omit outliers?
@AloneS @zepdrix @mathmate

Answer 1

Yes, if you can identify them.
Note: LSRL means least squares regression line.

Answer 2

How do I identify them again?

Answer 3

Is there an easier way to do it without going through the whole ISR thing?

Answer 4

IQR*

Answer 5

There are many ways to do it, depending on the recommended method.
If you teacher has shown you IQR method and nothing else, that's the one you should be using.

Answer 6

My teacher/lessons has taught us the IQR method and also outliers and influention points. However, while they did show example graphs of these, they did not explain exactly how to determine any other way to prove a data point is an outlier, not did they state to omit the outliers from the LSRL calculation. They just showed how the LSRL could differ based on outliers/influential points

Answer 7

nor*

Answer 8

Outliers is a slightly fuzzy subject. There are various methods for single-variable data sets. For multi-variable data, it seems that it still visual.
Example:
https://www.haesemathematics.com.au/media/W1siZiIsIjIwMTUvMDMvMTkvMm9hdTh1MXFndF9pYnNzbF8yXzIwLnBkZiJdXQ/ibssl-2_20.pdf
See page 596.

If your teacher is also vague about it, it's probably because there is no universal (and simple) method of identifying outliers, or it is outside of your course syllabus.
See if you can plot your points and determine if there are outliers. Do document what you did, even if you did not reject any data point. Hopefully your teacher will address this point in class after the corrections.

Answer 9

Okay thank you so much. :)

Answer 10

You're welcome. :)