Sample-data demonstration
This shows our AI testing process on public sample data. You can't upload your own data here.
From YourCloudGroup Model Manager.

Question: Which website visitors will buy
Sample data: Online Shoppers Purchasing Intention (UCI)
Attribution: Sakar, C. & Kastro, Y. (2018). Online Shoppers Purchasing Intention Dataset [Dataset]. UCI Machine Learning Repository. https://doi.org/10.24432/C5F88Q. CC BY 4.0. Modified: split into training and held-out rows, predictions added.
Predicts: Revenue (Classification)

Result
Its predictive value is STRONG, measured on held-out rows that took no part in choosing it.
Certification: CERTIFIED — AUDITABLE & REPEATABLE PREDICTIVE MODEL
Value verified: yes. The winning model demonstrated strong predictive value over the baseline on holdout rows that took no part in choosing it.
Winner: Random Forest, best of 12 models built
Trained on: 10,000 rows (the same 10,000-row sample of the 11,097 training rows for every model)

Certificate figures: the holdout half of the blind rows, 608 rows that took no part in choosing the winner
Balanced-accuracy lift: 52.4% (90% range 43.5-60.6%)
Balanced accuracy: 76.2% vs 50.0% for the baseline
Baseline: always predict the most common training class (False)

Compare figures: all 1,215 blind rows, including the half the winner was chosen on
Correct: 1,086 of 1,215
Confusion (rows = actual, columns = predicted): actual 0 (no purchase): 0 (no purchase) 970, 1 (purchase) 61; actual 1 (purchase): 0 (no purchase) 68, 1 (purchase) 116
The product marks these Compare figures as diagnostic: "These results do not establish predictive performance" (outcome timing was not declared for this run).

The predictions file
1,215 held-out rows, one per line: heldout_row, Administrative, Administrative_Duration, Informational, Informational_Duration, ProductRelated, ProductRelated_Duration, BounceRates, ExitRates, PageValues, SpecialDay, Month, OperatingSystems, Browser, Region, TrafficType, VisitorType, Weekend, actual_Revenue, predicted_Revenue, confidence.
heldout_row numbers the held-out rows 1 to N in the order the product tested them. It is not a row number from the original dataset: the split shuffled the rows.
Predicted classes are shown with the dataset's own labels, using the product's Compare encoding (0=False, 1=True).

Run: model a3bcb3f3-32a1-48da-85b2-516228137d42, tournament online_shoppers_20261028-210758, completed 2026-10-28T21:11:30Z
Dataset file sha256: b3055ee355f59134d851d32641183cb4a8b45def7124d2f50442a042f358e0d9