23  Predictive Modeling for TAT

23.1 Introduction

Predictive modeling enables proactive management by forecasting turnaround times and identifying at-risk cases before delays occur.

23.2 Load and Build Models

23.3 Model Performance

Regression Model (TAT Prediction): - Train R²: 0.004 - Test R²: 0.004 - Test RMSE: 124.91 hours

Classification Model (Delay Prediction): - Accuracy: 9.3%

Confusion Matrix: Delay Prediction

	Delayed	On-Time
Delayed	447	4501
On-Time	11	18

23.4 Actual vs Predicted

23.5 Feature Importance

23.6 Risk Stratification

Risk-Based Case Categorization

risk_category	n_cases	actual_delays	delay_rate
High Risk	4892	432	8.8
Low Risk	12	5	41.7
Medium Risk	73	21	28.8

Application: Use delay probability to prioritize cases for expedited processing.