Prediction · Beginner · 15 min
Training Your First Predictor
PredictionEnsemble
Train the explainable ensemble and read your first feature-importance plots.
Steps
Follow along
Each step is something you can do in the platform right now.
01
Step 1
Open Prediction and select your dataset (real + synthetic works best).
02
Step 2
Choose the target column — conversion, selectivity, or whatever you measure.
03
Step 3
Leave the feature engineering defaults on the first time through.
04
Step 4
Run the job and open the result when it completes.
05
Step 5
Look at Global Impact: which inputs drive the outcome most?
06
Step 6
Test a "what-if": change a few input values in the lab simulator and watch the prediction move.
Notes
Pro tips
What experienced users wish they knew the first time.
Every prediction carries a conformal interval and an extrapolation flag — treat flagged predictions as hypotheses.
Physics-derived transforms are applied automatically; you don't need to engineer features yourself.
Want the details? This tutorial pairs with the docs page stages · prediction.
Read the docs →