Brainstorm feature engineering ideas for a model
Use to generate a ranked list of candidate features from your raw columns and target.
You are an ML engineer doing feature engineering.
Prediction target: {{target}}
Available raw columns: {{raw_columns}}
Data grain: {{grain}}
Domain: {{domain}}
Model family I plan to use: {{model_family}}
Task: propose candidate features grouped by type:
- Aggregations and rolling windows.
- Ratios and interactions.
- Date and time decompositions.
- Categorical encodings appropriate for the model family.
- Lag and trend features (if temporal).
For each feature give: name, definition, the hypothesis for why it predicts the target, and a leakage risk note. Rank the full list by expected signal-to-effort. Flag any feature that risks target leakage in bold.Click the copy button in the top right of the block to grab the full prompt.
Replace each placeholder below with your own values before you run the prompt.
- {{target}}
- {{raw_columns}}
- {{grain}}
- {{domain}}
- {{model_family}}
Related prompts
You are a meticulous data analyst. Here is a sample of my raw dataset (header row plus a few rows): {{sample_rows}} Context: this data describes {{data_description}} and I plan to...
You are a senior Python data engineer. Write clean, reproducible pandas code to clean my dataset. Columns and their meaning: {{column_spec}} Known issues to handle: {{known_issues}...
Act as a data analyst guiding me through exploratory data analysis. Dataset summary: {{dataset_summary}} My goal / the question I care about: {{analysis_goal}} Produce an EDA plan...
You are a SQL expert writing for a {{sql_dialect}} database. Here is the relevant schema (tables, columns, types, keys): {{schema}} Question to answer: {{question}} Rules: - Use on...
You are a SQL reviewer. Explain and debug the query below. Dialect: {{sql_dialect}} What I expected it to return: {{expected_result}} What is actually wrong (if known): {{symptom}}...
You are a database performance specialist for {{sql_dialect}}. Slow query: {{query}} Context: - Approximate row counts of the main tables: {{table_sizes}} - Existing indexes: {{ind...
0 Comments
Loading discussion...