# Chapter 18: Model Building

924 learners taking this course

## Question 1 ~ Part 1 (Preview)

Length: 4 minutesAuthor: JasonComplexity: Standard

This is the most complex regression model we’ve seen so far. It includes squared variables (also known as second order), the regular first-order variables, and an interaction term. I discuss here how to write this out using the output as a guide.

## Question 1 ~ Part 2-3

Length: 24 minutesAuthor: JasonComplexity: Standard

More fill in the blanks! Hooray! The methods used to find the missing values are the same as we used in Chapter 17 with little to no variation - so this section shouldn't hold any surprises for you.

## Question 1 ~ Part 4-5

Length: 19 minutesAuthor: JasonComplexity: Hard

PART 4 asks us to test the overall validity of the model, or stated more simply do the independent variables (taken as a group) help to predict the annual sales. In PART 5 we are given the values of the independent variables and must provide an actual estimation of sales. This is a question that MANY students get wrong.

## Question 2a

Length: 23 minutesAuthor: JasonComplexity: Standard

How to create indicator variables when you are given raw nominal data. I go into detail about the binary (yes/no) nature of these new variables as well as the importance of having a 'missing' category.

## Question 2bc

Length: 9 minutesAuthor: JasonComplexity: Standard

Part (b) asks us for the POPULATION regression model, while part (c) is looking for the SAMPLE regression model. Learn what the difference is - and how to write both when working with indicator (dummy) variables.

## Question 2d

Length: 13 minutesAuthor: JasonComplexity: Standard

Working with the coefficients of indicator variables can be confusing. The test of slope requires a new interpretation. I show you how the new wording connects with what you learned in simple and multiple linear regression.

## Question 2e

Length: 13 minutesAuthor: JasonComplexity: Easy

Just like with the test of slope, interpreting the coefficient of an indicator variable requires new wording to get it right. Remember - indicator variables are not simply nominal variables, they are BINARY nominal variables.

## Question 2f

Length: 2 minutesAuthor: JasonComplexity: Easy

Here we complete the test of the hypotheses created in part (d) using a 5% significance level. These tests will tell us whether or not there is a difference between the category represented by an indicator variable and the 'missing' category discussed in part (a)

## Multiple Choice

Author: Jason

Test your understanding of Chapter 18: Model Building. Each question is accompanied by a mini-video lecture showing you how I decided which solution was the correct one.