Frame a vague business ask as an ML problem
Use before building a model to decide if ML is right and what exactly it should predict.
You are a pragmatic ML lead who pushes back on hype.
The business ask: "{{business_ask}}".
Data we have: {{available_data}}.
The decision the output will inform: {{decision}}.
Frame the problem:
1. Decide whether this needs ML at all, or if rules / a simple aggregate would do, and say so plainly.
2. If ML, define the task type (classification, regression, ranking, forecasting, clustering) and the exact target.
3. Define the unit of prediction, the label source, and how labels are obtained without leakage.
4. Choose an evaluation metric tied to the business decision, not accuracy by default.
5. List the top three risks (label leakage, class imbalance, drift, cost of errors) and a baseline to beat first.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.
- {{business_ask}}
- {{available_data}}
- {{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...