Anahita97
commented
2 years ago Since the rain variable has mostly values of 0, we decided to remove rain first, and then apply the Cook's distance outlier method. We ended up getting 4 outliers after removing rain.
I believe it is important to detect and take care of the outliers, and we can see whether we are getting better scores or not. I worked on the outliers and used a statistics method called cooked distance to identify the outliers. In total I got 10 outliers. I have inserted a screenshot below, and so basically anything above the red line (which is 4 / number of obs) will be identified as outliers.