STPP — Stochastic Thermal Property Predictor

Model Pipeline

Step 1 Upload CSV

→

Step 2 Train Model

→

Step 3 Get Predictions

3-STEP WORKFLOW — EDA · Hyperparameters · Batch or single-row output

Supported Algorithms

3 Models

Random Forest • Gaussian Process • SGD Bootstrap

Uncertainty Formula

μ ± 1.96σ

Approximate 95% prediction interval

Prediction Output Columns

CSV Export

            prediction_mean
          

prediction_lower_95 • prediction_upper_95

Calibration Report Levels

50% 80% 90% 95%

Nominal vs empirical coverage chart generated after every training run

Algorithm

Ensemble of decision trees — fast, scalable, robust uncertainty via tree-spread standard deviation.

Target Column

Auto-detected from your uploaded CSV — any numeric column.

Best For Any dataset size, especially > 500 rows

Next Step Open app → Data tab → upload CSV → Train tab → select features