Module 10: Prediction and PLSpredict

Purpose

SmartPLS models are often used for prediction. This module introduces how to assess predictive performance, especially with PLSpredict for PLS-SEM.

Learning Objectives

After this module, learners should be able to:

• Explain why prediction assessment is different from significance testing.
• Interpret Q2_predict.
• Compare RMSE and MAE values.
• Report predictive performance responsibly.

Prediction vs Explanation

Significant paths do not automatically mean strong prediction. Predictive assessment asks whether the model predicts new or holdout cases well.

PLSpredict

PLSpredict uses cross-validation to generate prediction errors for indicators and constructs. SmartPLS documentation describes RMSE, MAE, and MAPE for manifest variables and RMSE/MAE for latent variables.

Q2 Predict

If:

Q2_predict > 0

then the model's prediction error is smaller than using a simple mean benchmark.

LM Benchmark

SmartPLS can compare PLS-SEM prediction errors with a linear regression model benchmark for manifest variables. Lower RMSE or MAE indicates better predictive performance.

Reporting Template

Predictive performance was assessed using PLSpredict. The Q2_predict values were positive for ..., indicating that the model outperformed the mean benchmark. The PLS-SEM model produced lower RMSE/MAE than the linear model benchmark for ... indicators, suggesting [low/medium/high] predictive performance.

Practice

1. Run PLSpredict if your model uses PLS-SEM.
2. Export Q2_predict results.
3. Compare RMSE and MAE with the LM benchmark.
4. Write one prediction assessment paragraph.

Next Module

Continue to Module 11: Common Reviewer Problems.