I am using the yahoo stock price dataset from kaggle site to practice with the TS tools. When I use the ARIMA or ETS tool on the training dataset(leaving out the last year for validation), both models seem to be heavily overfitting. I tried to calibrate ARIMA with different parameters but could not find a way to reduce overfitting on the training data. Does anyone have any suggestions regarding how to avoid overfitting in these cases? Thank you in advance,
Hi @alkan
Could you elaborate/share how you're customizing the parameters? Also if you can share your data/workflow it's very helpful in these situations for others to replicate.
Thank you but I figured out the issue.