Which of the following methods help when a model suffers from high variance? a. Increase training data. b. Increase model size. c. Decrease the amount of regularization. d. Perform feature selection.

Is the retrospective test an appropriate approach to test the model?

If a model is prescriptive, what makes it so?

Three classifiers are to be benchmarked. To this end, using the same data, the classifiers were trained and the following table shows the validation results obtained with n = 16 observations. 1 1 0 2 0 3 1 4 1 5 1 6 1 7 0 8 9 10 11 12 13 14 15 16 OTTOTOO 0 1 1 1 1 0 ZOOOoooo Hooo o Ytrue Y1 Y2 Y3 1 0 0 0 1 0 1 0 1 1 0 0 1 1 0 1 0 0 0 0 0 0 0 1 1 0 0 1 0 0 1 1 0 1 0 0 0 1 1 0 11O 1 1 1 1 0001 Match the classifiers with the performance measures. Accuracy and Error rate for Y3 Choose... Accuracy and Error rate for Y2 Choose... TPR and FPR for Y1 Choose...

Question # 2: Consider the following data Classification Model where YACT is your actual observation and YPRED is the model prediction value. You have to use the data and find CONFUSION MATRIX, and using confusion matrix compute the value of the following errors: Precision value of each class Recall value of each class F-measure value of each class Model Accuracy Model Precision value Model Recall value Model F-measure value YACT R G B B R G R R G G B B R R G B R B G R B G B R R B B G G G YPRED R R R B B B R R G B B R R G G G B B R R B G B R B G B G R R

Which ensemble technique, varies features to train the models? which ensemble technique, varies training data?

First, perform the following tasks: • Make a linear regression model with all the features in the dataset. Use train_test_split to keep 20% of the data for testing. • Use your model to predict values for test set and print the predictions for the first 10 instances of the test data and compare them with actual values. • Print the coefficient values and their corresponding feature name (e.g. age 43, bmi 200, .) • Note that you can access feature_names from diabetes dataset directly • Calculate training-MSE, testing-MSE, and R-squared value. Compare the two models. Did using all available features improve the performance? In [ ]: # Your code goes here In [ ]: # Your code goes here

A model's correctness can only be evaluated by its performance on test data. describe in depth; provide more information? Explain?

Use python Machine learning to answer the following questions. 1. What are training, validation, and testing data for? Why did we use validation data to find the best alpha? Can we use test data to find the best alpha? 2. What is the difference between KFold and train_test_split? What is the advantages and disadvanteges of k-fold cross validation?

3 - You are asked to choose one of the following two methods to measure the classification performance of your model: i) 10-fold cross validation or ii) split the data into 70%-30% for model training-testing and repeat it 10 times. Write the number of frauds that you will use to test model performance in each of the two methods and argue which one is more appropriate. - You are not satisfied with the classification performance of your model and take some steps back to understand how can you improve it. By looking at basic descriptive statistics of the data, you notice that features have different order of magnitude, i.e. some are in tens while others are in millions. Mention one method to deal with such a problem. -The classification performance of your model is still poor after all you have done. Mention what is the biggest challenge in the context of fraud detection and mention some solutions to attempt to solve such a problem.