Welcome to the World of Education.

Discriminant Analysis is somewhat similar to regression analysis. To predict dependent variables in both techniques, we can use dependent and / or some independent variables. In this, the dependent variable is categorical, not metric. This can be used to classify people or objects into two or more groups based on some knowledge of their characteristics.

Examples

1. On the basis on some demographic or psycho graphic characteristics of the consumers, we can predict or identify preferences to the brand name or change in brand preferences.

2. Scores in selection test can be a criteria to select MBA students for the post of sales person or any other post.

The discriminant analysis may be depend on or require data on all the independent variables and the categories to which each object belongs.

For example, selection of salesperson – whether each salesperson was selected or rejected for a given set of scores in selection tests

Then discriminant model built and tested usefulness on the basis of

1. Its significance

Low value of Wilk’s Lambda : High significance

The F test should show a p value of less than .05

2. Which variables are relatively better at predicting the dependent?

This can be judged by looking at the standardized coefficients of the independent variables

The larger the absolute value of the standardized coefficient, the better the predictive  or explanatory power of the variable.

3. The number of data points from the original dataset the model classifies correctly.

Closer to 100 is better

Share This