I am working on building a Random Forest Regression model for predicting ETA. I am saving the model in pickle format by using pickle package. I have also used joblib to save the model. But the size of file is really large (more than 100 GB). I would like to ask the data science experts that is it the correct format to save the model or is there any other efficient method to do so? Any insights on this will be appreciated.
Asked
Active
Viewed 276 times
0
-
how many n_estimators (trees) do you have in your forest? – Gabriel M Oct 10 '18 at 14:54
-
I have 400 estimators – Vikika Oct 10 '18 at 19:08
1 Answers
0
Aim of saving complete model is modifying the model in the future. If you are not planning to modify your model, you can just save the weights and use them for the prediction. This will save a huge space for you.
murat yalçın
- 709
- 7
- 10