Discriminant Analysis Example

Discriminant Analysis Example

Problem Statement

Use the discriminant analysis model to predict the values for gear based on the other variables. 

How to perform analysis

Step 1: Open Sigma Magic
  1. Click on the Sigma Magic button on the Excel toolbar.
  2. Click on the New button to create a new project.
Step 2: Add the analysis template
  1. Click on the Tool Wizard to add the analysis template.
  2. Click on Analytics and then Discriminant Analysis.

Step 3: Specify analysis options
A new worksheet will be added to your workbook. Analysis Setup will be automatically opened, in the setup tab specify the survey results.
 

Click on Data to specify the data required for this analysis.   


Click on the Train the software will let you pick the options for training the given model. Training is a step where we split the data into groups a train data set and a test data set.


Click on the Tuning to identifying the best set of hyperparameters that gives the best fit for the given model.


Click the Verify tab to ensure all the inputs are okay and shown in a green checkmark. 


Step 4: Generate analysis result
Click OK and then click Compute Outputs to get the final results.


Interpretation of Results

  • The Linear Discriminant Analysis (LDA) accuracy is reported as 75.7%, which means the model correctly classifies approximately 3 out of 4 observations.
  • Accuracy can be improved by adding more relevant predictors, removing multicollinearity, or balancing the dataset.
  • The discriminant function coefficients indicate the relative importance of each variable in differentiating between classes.
  • The group centroids provide insights into how well-separated the categories are. The greater the centroid difference, the better the separation.
  • The confusion matrix (not fully visible here) likely shows misclassified observations.
  • If misclassification is high for specific groups, consider transforming data or using Quadratic Discriminant Analysis (QDA).
  • The Mahalanobis distances suggest how far each observation is from the group centroid.
  • Higher distances indicate observations that might be misclassified or outliers.
  • The analysis assumes equal covariance matrices for LDA.
  • If the assumption is violated, Quadratic Discriminant Analysis (QDA) or Regularized Discriminant Analysis (RDA) might be more appropriate.
  • The conclusion states that the analysis meets assumptions but suggests room for improvement.
  • Consider cross-validation, feature selection, or alternative models (e.g., logistic regression, decision trees) for better classification.

    • Related Articles

    • Discriminant Analysis Overview

      Discriminant Analysis is a statistical technique used for classifying observations into predefined groups based on predictor variables. It is commonly applied in cases where the dependent variable is categorical, and the goal is to determine which ...

    • Correspondence Analysis Example

      Problem Statement We have collected data on several people's eye and hair color. Use correspondence analysis to determine if there is any correspondence between the two. How to perform analysis Step 1: Open Sigma Magic Click on the Sigma Magic button ...

    • SMED Analysis Example

      Problem Statement Create a SMED analysis for reducing the changeover time for planes between two flights. # Description 1 Check if all personnel are present 2 Security check 3 Gather all tools & supplies 4 Cleanup plane 5 Refurbish plane 6 Clean ...

    • Waste Analysis Example

      Problem Statement Select a process and create the waste analysis for each step in the process. How to perform Analysis Step 1: Open Sigma Magic Click on the Sigma Magic button on the Excel toolbar. Click on the New button to create a new project. ...

    • PEST Analysis Example

      Problem Statement Create a PEST model for one of the products that is launched by your company. Investigate all the opportunities and threats for this product using the PEST model. How to perform analysis Step 1: Open Sigma Magic Click on the Sigma ...