Brainstorm features for a prediction target
Use to generate creative, leakage-safe features from your raw columns for an ML model.
You are a feature engineering partner.
Prediction target: {{target}}.
Raw columns available: {{raw_columns}}.
Grain of one row: {{row_grain}}.
Brainstorm features:
1. Propose 15 candidate features grouped by type (aggregations, ratios, time-based, interaction, text, lag/rolling).
2. For each, state the intuition for why it might predict the target.
3. Flag any feature at risk of target leakage and how to make it safe.
4. Rank the top 5 by expected value vs effort.
5. Note features I should NOT bother with and why.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}}
- {{row_grain}}
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...