Model Performance Dashboard

Comparative evaluation of 6 trained classifiers on the water potability dataset

Best: Random Forest
Trained Jun 7, 2026

Best Model Accuracy

83.5%

Random Forest

+3.4% vs Logistic Regression

ROC-AUC Score

0.871

Random Forest

Best discriminator

F1 Score — Safe Class

0.854

Precision-Recall balance

Correctly identifies safe water

F1 Score — Unsafe Class

0.806

Critical — false negatives

Monitor false negative rate

Model Comparison

ROC Curves — All Models

True Positive Rate vs False Positive Rate

Random Forest
0.871
Extra Trees
0.858
Gradient Boosting
0.846
SVM
0.821
Log. Regression
0.734
KNN
0.759

Full Model Comparison

6 models · 5-fold CV
ModelCV ± StdKey ParamsStatus

Random Forest

★ Best model

83.47%87.12%85.41%80.63%82.01%82.0%±1.24
n_estimators=400max_depth=20
Best

Extra Trees

82.19%85.84%84.01%79.32%80.89%80.9%±1.48
n_estimators=600max_depth=20
Good

Gradient Boosting

81.03%84.61%82.67%78.41%79.78%79.8%±1.67
n_estimators=200learning_rate=0.1
Good

SVM

78.42%82.13%79.94%76.12%77.01%77.0%±1.93
C=10kernel=rbf
Baseline

KNN

71.26%75.89%73.41%67.82%70.14%70.1%±2.41
n_neighbors=7weights=distance
Baseline

Logistic Regression

69.14%73.42%72.01%64.11%68.23%68.2%±2.18
C=1penalty=l2
Baseline