Split the entire dataset into three parts, One is a train, two is validation, and then a test set. Build a DNN model with train data, tune hyper-parameters with validation data, and finally evaluate performance on the test data.
computer science
Description
1. Describe at least two ways of ensembling together DNNs and RFs.
Take any dataset from Kaggle and
(A) train an RF model,
(B) train a DNN with the following steps-
Steps-
(i) pre-process the data by scaling/standardizing the columns
(ii) Split the entire dataset into three parts, One is a train, two is validation, and then a test set. Build a DNN model with train data, tune hyper-parameters with validation data, and finally evaluate performance on the test data.
(iii) Make Epoch versus train set accuracy, and validation set accuracy
(iv) Report results using nice ROC curves, report AUC values.
(C) a hybrid DNN and RF model.
Provide a detailed model and result comparisons.
