How to minimize volatility in dataset?

Hi guys,

We are currently struggling with a bit of an issue with regards to multiple regression.
It concerns how to deal with substantial volatility in a dataset, which may affect least squares regression.

The exact question goes like this:
"The company has noticed that the dataset contains reasonably substantial volatility that may affect least squares regression. Explain what tools you would use to provide a more robust form of regression".

My best suggestions to minimize volatility would be to:
1) Take the square-root of each observation value
2) Increase sample size

However, I'm honestly not sure if this is in alignment with the question and/or whether or not it would contribute to create a more "robust" form of regression?

Any advice is highly appreciated! :wave:


Fortran must die
Variation in a data set is desirable I always thought. Certainly concerns are constantly raised that there is too little variation which can influence statistical tests. I have never heard that too much variation is an issue.