TNZL JC311 Lab
Rathna Products
TNZL JC311 Lab  ·  SRM University–AP

Predict Wastewater Treatment Efficiency

Machine-learning tool for researchers and engineers — predict TOC, COD, and UV Absorbance removal efficiency for electrocoagulation (ECP) and photo-assisted electrocoagulation (ECP+PR) systems.

87Training Observations
9Published Studies
3Target Variables
5ML Models
Start Prediction ↓

Removal Efficiency Predictor

Enter your experimental conditions below. The tool uses an Extra Trees ensemble trained on peer-reviewed literature data.

Process & Material Parameters

ECP+PR uses a UV lamp coupled with electrocoagulation

Operating Conditions

Typical range: 1 – 65 mA/cm²
Solution pH before treatment
0 min 60+ min

Prediction Results

🔬

Enter parameters and click Predict to see results

Model Performance

Validated using Group K-Fold cross-validation (grouped by Study_ID to prevent data leakage between studies).

TOC Removal
OOF R² 0.374
Full-data R² 0.975
Training Rows 61
Moderate generalization
COD Removal
OOF R² 0.095
Full-data R² 0.915
Training Rows 48
Indicative — high uncertainty
UV Absorbance Removal
OOF R² 0.792
Full-data R² 1.000
Training Rows 28
Most reliable prediction
⚠ Important: Predictions are ML estimates based on 87 training observations from 9 published studies. The OOF R² reflects true cross-study generalization. Always validate predictions experimentally before making operational decisions.

About This Tool

📊

Dataset

87 observations curated from 9 peer-reviewed studies + in-house experimental data. Covers ECP and ECP+PR systems across municipal wastewater, greywater, textile effluent, landfill leachate, and industrial wastewater.

🤖

Algorithm

Extra Trees Regressor ensemble (300 estimators). Five models were trained and compared: Random Forest, Gradient Boosting, Extra Trees, XGBoost, and Voting Ensemble. Extra Trees showed best OOF performance for UV removal.

✅

Validation

Group K-Fold cross-validation (5-fold, grouped by Study_ID) prevents data leakage between publications. Both OOF R² (honest generalization) and full-data R² (training fit) are reported.

🔬

Key Predictors

Feature importance analysis identified Treatment Time and Current Density as the most influential predictors, followed by Process Type (ECP vs ECP+PR), Anode Material, and Initial pH.