MKT 3340 Study Guide - Midterm Guide: Logistic Regression, Conjoint Analysis, Logit

• School: UT-Dallas
• Department: Marketing Management
• Course: MKT 3340
• Professor: Prasad Ashutosh

Document Summary

Residuals difference between predicted value of with observed (actual) value of y for every x. This makes the estimates from a multiple regression model unstable" in that t-tests for them may show insignificant even though the f-test shows the regression model is significant. A small change in data can affect them. Can be discovered by examining the correlation among x variables. Possible solution is to pick only one of two highly correlated values. R open source, desktop to enterprise level. Mathematica, maple, matlab proprietary, mathematical software. Regression: relationship between a metric variable (y) and a function of one or more metric variables, y = alpha + beta x + epsilon. Dummy variable regression: the x variables in the regression include categorical variables. Logit model: when the y variable in the regression is a categorical variable. Logit is used to explain choice or membership of a category.