description.txt

In this project, we used sklearn's RandomForestRegressor to predict a train and validaton error. 
The features we used were: training and test accuracy for the first 50 epochs, and training and test loss, which were all combined into a single vector per sample, with the labels being the final train and validation error provided.
We achieved a ~95.9% accuracy on the public half of the dataset.

Here are some statistics from our model's training:

Train Error
-----------
Mean Absolute Error: 0.010916260139055764
Mean Squared Error: 0.00028738106109099903
Root Mean Squared Error: 0.016952317277912157

Validation Error
----------------
Mean Absolute Error: 0.008713028643137517
Mean Squared Error: 0.00017969941231493143
Root Mean Squared Error: 0.013405200942728589