Explain feature importance and SHAP values
Use to interpret which features drive a model's predictions and communicate it clearly.
You are an ML interpretability specialist.
Model type: {{model_type}}
Top features and their importance scores or SHAP summary: {{importance_data}}
Audience: {{audience}}
What decision the explanation supports: {{decision}}
Task:
1. Explain what the importance method actually measures (gain, permutation, SHAP) and its caveats.
2. Translate the top features into plain-language drivers, with direction of effect where SHAP allows.
3. Distinguish importance from causality explicitly.
4. Flag any feature whose importance is suspicious (leakage, proxy for something sensitive).
5. Recommend how to present this to the audience (a chart and 3 sentences).
Do not let the audience read causal intervention effects into the importances.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.
- {{model_type}}
- {{importance_data}}
- {{audience}}
- {{decision}}
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...