Create intermediate Works with: Any AI

Feature Engineering Brainstorm

Table output

Create features from raw data thoughtfully.

The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.

Brainstorm features for a predictive model. Prediction target and decision time: [TARGET_AND_TIMING]. Raw tables and columns: [RAW_DATA]. Domain knowledge I have: [DOMAIN_KNOWLEDGE].

Produce a table of candidate features: Feature | How to compute | Why it might help | Availability at prediction time (yes or no) | Leakage risk | Effort. Organize by type: counts and sums over windows (last 7, 30, 90 days), recency and frequency, ratios and rates, trends (change versus the previous period), categorical encodings (target encoding only with cross-validation), text length and keyword flags, calendar features (day of week, month, holiday), interactions suggested by domain logic, and missingness indicators.
Add: the code for three of them using groupby and rolling windows with a cutoff date so no future data is used; a method to test whether a feature helps (ablation on validation); a warning against creating thousands of features without a hypothesis; and a list of features that could encode protected characteristics indirectly and need review [VERIFY in [JURISDICTION]].

Fill in before use

Optional: type your details and the prompt above updates. Nothing is saved or sent.

Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.

• feature-engineering