You are building a ML model to predict whether houses for sale in a particular neighbour- hood will be sold in the next month or not. You have a dataset of 10,000 houses, their sale prices, date of purchase and other relevant attributes. (a) What will be your training-test data set split? (6) You get 20,000 new samples as test data for your model evaluation. You notice that your model correctly predicts Yes 70% of the time and correctly predicts No. 30% of the time. If there are 18,000 actual YES samples in the dataset, compute the confusion matrix, Type I errors and Type II errors and accuracy. (e) You tune hyper-parameters for your model and get the following ROC curve. Identify the parameter configuration (1, 2, 3 or 4) will you choose and justify the choice. [1] 2U 100% 3 80% 60% True ? Positive Rate (Sensitivity) 40% -1 20% 0% 0% 100% 20% 40% 60% 80% False Positive Rate (1-Specificity)