I'm fairly new with ML and could use your collective advice here.
I have a regression problem (outcome is price) and I am using a random forest estimator model, based on sklearn.ensemble.RandomForestRegressor.
The results I'm getting are shown in the scatter plot here:
In the training dataset (blue), the model is not over fit and performs ok. In the test dataset (orange), the model seems to overestimate the low actual values and underestimate the high actual values.
I'm trying to interpret this behavior and extract some ideas on how to tune the model or adjust the pre-processing (outliers, nulls etc).
Given that I'm fairly new to this, this may be a naive or too broad question but any suggestions are appreciated.
Thank you!