TK Software is a firm that sells software and games. It recently acquired several new products and invested in a marketing campaign that reached 2000 of its previous customers with the expanded offerings. Based on the responses to this campaign, it would like to develop a model for predicting the amount of spending of a purchasing customer. The table below contains variables available for modeling.

FREQ Number of transactions in the preceding year
LAST_UPDATE Number of days since last update to customer record
FIRST_UPDATE Number of days since the first update to customer record
WEB Whether the customer purchased by Web order at least once
GENDER Male or female
ADDRESS_RES Whether it is a residential address
ADDRESS_US Whether it is a U. S. address
SPENDING (outcome Amount spent by customer in test campaign ($)

a.) Describe how categorical variables like GENDER can be included in a regression analysis.

[Insert response.]

b.) An analyst is interested in consistency of the model parameters to different samples and splits the data into multiple training and validation sets. The table below shows the parameters obtained when the model is fit to the first and the second training set.

Coefficients:
1st Dataset 2nd Dataset
Intercept 15.630 125.628
FREQ -11.369 -172.851
LAST_UPDATE 95.955 195.960
FIRST_UPDATE -0.012 0.188
WEB 0.001 -0.251
GENDER 13.856 -63.385
ADDRESS_RES 3.933 277.556
ADDRESS_US -97.968 -162.767

What inconsistencies do you notice? What is the most likely cause of these inconsistencies? What would you recommend as next steps for the analyst in order to obtain a multiple regression model with parameter values that would remain relatively consistent when fit to different subsets of the data? (i. e. how would you rectify the identified issue?)