Machine Learning
Model Performance
Trained on 79 sessions · 23 positive labels · Small dataset — treat metrics as directional
LR Accuracy
75.0%
AUC 0.545
RF Accuracy
81.3%
AUC 0.746
Best Model
RF
Random Forest
Best AUC
0.746
Higher = better separation
Logistic Regression — Confusion Matrix
Predicted No
Predicted Yes
Actual No
10
True Neg
1
False Pos
Actual Yes
3
False Neg
2
True Pos
16 test sessions total
Random Forest — Confusion Matrix
Predicted No
Predicted Yes
Actual No
10
True Neg
1
False Pos
Actual Yes
2
False Neg
3
True Pos
16 test sessions total
ROC Curve
LR AUC: 0.545RF AUC: 0.746
Feature Importance
Which AR behaviours most predict cart conversion — Random Forest
Gold = highest impact feature
Live Predictor
Adjust session behaviour to predict cart conversion likelihood
0
0
0
0
0
0
Conversion Probability
31.6%
Unlikely to convert
Computed from Logistic Regression coefficients stored at last training run.
Training History
| Run | Date | Sessions | Cart Events | LR Accuracy | RF Accuracy |
|---|---|---|---|---|---|
| #2Latest | 07 Jun 2026, 19:22 | 79 | 23 | 75.0% | 81.3% |
| #1 | 07 Jun 2026, 15:53 | 69 | 17 | 85.7% | 85.7% |