Question

8) Two interns, Ajax and Sarpedon, are attempting to model what makes a band successful. Ajax creates a regular model, while Sarpedon takes Ajax’s model, adds more variables, and creates an overfit model. Which of the following statements is true? (The p-values mentioned below refer to a test where H0 = “the stupid line” is correct.)

a) Compared to Ajax’s model, Sarpedon’s model has a higher p-value, and will predict future bands’ success more accurately.
b) Compared to Ajax’s model, Sarpedon’s model has a higher p-value, and will predict future bands’ success less accurately.

c) Compared to Ajax’s model, Sarpedon’s model has a lower p-value, and will predict future bands’ success more accurately.
d) Compared to Ajax’s model, Sarpedon’s model has a lower p-value, and will predict future bands’ success less accurately.

Answer 1

8

a) Compared to Ajax's model, Sarpedon's model has a higher p-value, and will predict future bands' success more accurately will not be true because the model is an overfit model that means that while training the model it performs more accurately but on testing the model, it performs poorly compared to Ajax's model i.e., it predicts bands' success less accurately.

b)Compared to Ajax's model, Sarpedon's model has a higher p-value, and will predict the future bands' success less accurately is not true because the model is an overfit model and performs greatly when the model is being trained thus having a lower p-value but performs poorly when the model is being tested.

c)Compared to Ajax's model, Sarpedon's model has a lower p-value, and will predict the future bands' success more accurately will not be true because first of all, overfitting a model results in a higher p-value compared to the original model and secondly, the overfit model predicts the future bands' success more accurately on training the model but actually performs poorly on testing the model.

d)Compared to Ajax's model, Sarpedon's model has a lower p-value, and will predict the future bands' success less accurately will be true because overfitting a model decreases the p-value compared to the original model and thus will not be able to predict more accurately than the original model.

Training a model means to perform various statistical algorithms or machine learning algorithms on the dataset where the independent variable values are given along withthe dependent variables. Testing the model means based on the predictions obtained while training the model we use it on another dataset that contains only the independent variable to obtain the final result.