Methodologies

• Original datasets were chaotic and had to be pre-processed first.
• Then, the datasets were divided into a train set, a validation set, and a test set.
• For the Round Hill dataset, experiment numbers 14,15,26,32,33 were used as the test set, and the remaining 34 experiments were used as train and validation sets.
• For the Prairie Grass dataset, experiment numbers 19,27,37,57,67 were used as the test set, and the remaining 63 experiments were used as train and validation sets.
• All the features were normalized before use.
• Various hyper-parameters of the models were tuned by testing model’s performance on the validation set.
• Finally the models were tested on the hold-out test set.