Best AI Prompts for Data Analysis & Science

Write a SQL query for the following task: Task: [describe what you need, e.g. find all users who signed up in the last 30 days and made at least one purchase] Relevant tables and columns: [describe your schema or paste CREATE TABLE statements] Database: [e.g. PostgreSQL, MySQL, BigQuery] Return the query with comments explaining each key step.

This SQL query is running slowly. Help me optimize it. Query: [paste query here] Database: [e.g. PostgreSQL, MySQL] Table sizes: [e.g. orders has 50M rows, users has 2M rows] Existing indexes: [list them or paste EXPLAIN output] Identify the bottlenecks, suggest index changes, and rewrite the query if needed. Explain each optimization.

Explain this SQL query in plain English. Walk through it step by step — what each clause does, what data it produces, and what the final result represents: [paste query here]

Design a database schema for the following use case: Use case: [describe your application or feature, e.g. a multi-tenant SaaS with projects, tasks, and user roles] Database: [e.g. PostgreSQL] Include: table names, columns with types and constraints, primary/foreign keys, indexes, and an explanation of your design decisions and any trade-offs.

I have a dataset and want to perform exploratory data analysis. Create a step-by-step EDA plan. Dataset description: [describe columns, data types, approximate size, and domain, e.g. e-commerce orders with customer ID, product, price, date] Business question I am trying to answer: [describe it] Include: data quality checks, univariate analysis, relationships to explore, visualizations to create, and potential red flags to watch for.

Help me choose and interpret the right statistical test for my analysis. What I am comparing: [e.g. conversion rates between two user groups] Data type: [e.g. binary outcomes, continuous measurements] Sample sizes: [e.g. Group A: 1,200, Group B: 1,400] Current results: [e.g. Group A: 4.2%, Group B: 5.1%] Recommend the appropriate test, state the assumptions, show the calculation or code, and explain how to interpret the p-value and confidence interval in plain language.

Help me design and interpret a cohort analysis. Goal: [e.g. measure 30/60/90-day retention by signup month] Available data: [describe tables and columns, e.g. users table with signup_date, events table with user_id and event_date] Database / tool: [e.g. BigQuery, pandas] Write the query or code to build the cohort table, explain how to read the output, and flag any interpretation pitfalls.

I am seeing a trend in my data and need help interpreting it. Metric: [e.g. weekly active users] What I observe: [e.g. 15% decline over the past 6 weeks after steady growth] Context: [any known events, product changes, seasonality, or external factors] Help me: identify likely causes, rule out data artifacts, suggest additional analyses to confirm the root cause, and recommend next steps.

Recommend the best chart type for my use case and explain why. What I want to show: [e.g. how revenue breaks down by product category and changes month over month] Audience: [e.g. executive team in a quarterly business review] Data shape: [e.g. 12 months x 6 categories, values range from $10K to $2M] Tool: [e.g. Tableau, matplotlib, Looker Studio] Suggest 2-3 options, rank them, and explain the trade-offs.

Help me design an effective dashboard layout. Purpose: [e.g. daily operations monitoring for a support team] Key metrics to show: [list them, e.g. open tickets, resolution time, CSAT score, ticket volume by channel] Audience and how they use it: [e.g. team leads check it at the start of each shift] Tool: [e.g. Metabase, Tableau, Grafana] Propose a layout with sections, chart types for each metric, and UX best practices for at-a-glance readability.

Help me turn this data into a clear narrative for a presentation. Key findings: [list 3-5 main insights from your analysis] Audience: [e.g. non-technical stakeholders, C-suite] Goal of the presentation: [e.g. get approval to invest in retention improvements] Write a narrative arc that leads with the most important insight, supports it with data points, and ends with a clear recommendation. Keep it concise and free of jargon.

Help me write annotations for this chart to make it self-explanatory. Chart type: [e.g. line chart] What it shows: [e.g. monthly revenue for the past 2 years] Key moments to call out: [e.g. product launch in March, pricing change in August, a dip in December] Audience: [e.g. investors in a pitch deck] Write 3-5 short, punchy annotation labels and a one-sentence chart title and subtitle.

Generate a comprehensive data cleaning checklist for my dataset. Dataset description: [describe columns, data types, source, and domain] Known issues: [list any problems you already see, e.g. duplicate rows, inconsistent date formats, missing values in key columns] Downstream use: [e.g. feeding a machine learning model, building a dashboard] Organize the checklist by priority and include the specific checks to run and how to fix each issue.

Help me decide how to handle missing values in my dataset. Column with missing values: [column name and what it represents] Percentage missing: [e.g. 18%] Data type: [e.g. numeric, categorical, datetime] How the column is used downstream: [e.g. it is a feature in a regression model] Compare the options (drop rows, impute with mean/median/mode, model-based imputation, flag as a separate category) and recommend the best approach with reasoning.

Help me detect and handle outliers in my dataset. Column: [column name and what it represents] Data type: [numeric / datetime] Sample statistics: [e.g. min: 0, max: 99,000, mean: 450, median: 380, std: 1,200] Context: [e.g. this is order value in USD; very high values could be legitimate bulk orders or data errors] Suggest appropriate detection methods (IQR, z-score, isolation forest, etc.), provide code in [Python/SQL], and advise on how to treat confirmed outliers.

Help me standardize inconsistent values in this column. Column name: [e.g. country] Sample of messy values: [e.g. 'USA', 'United States', 'US', 'u.s.a', 'America'] Desired output format: [e.g. ISO 3166-1 alpha-2 country codes] Tool: [e.g. Python pandas, SQL, dbt] Write the code to map and standardize these values, handle edge cases, and flag any values that cannot be mapped automatically.

Write a pandas data pipeline for the following task. Input: [describe the data source, e.g. a CSV file with columns: user_id, event_type, timestamp, revenue] Transformations needed: 1. [e.g. filter to purchase events only] 2. [e.g. aggregate revenue by user per month] 3. [e.g. calculate month-over-month growth rate] Output: [describe desired result, e.g. a DataFrame with one row per user per month] Write clean, readable code with comments. Use method chaining where it improves clarity.

Write Python code to create the following visualization. Chart type: [e.g. heatmap, grouped bar chart, time series with confidence interval] Data: [describe the DataFrame structure, e.g. df has columns: month, category, value] Style requirements: [e.g. dark background, colorblind-friendly palette, no chart borders] Output: [e.g. save as PNG at 300 DPI] Use matplotlib and/or seaborn. Add proper title, axis labels, and legend. Make it publication-ready.

Suggest feature engineering ideas for my machine learning dataset. Prediction target: [e.g. whether a user will churn in the next 30 days] Available raw features: [list them with types, e.g. signup_date (date), last_login (date), num_sessions (int), plan_type (categorical)] Data granularity: [e.g. one row per user] Suggest 10-15 derived features, explain the intuition behind each, and provide pandas code to create them.

Write a Python data validation script for my dataset. Dataset: [describe columns, types, and source] Validation rules to enforce: 1. [e.g. user_id must be non-null and unique] 2. [e.g. event_date must be between 2020-01-01 and today] 3. [e.g. revenue must be >= 0] 4. [e.g. status must be one of: active, inactive, pending] The script should print a summary of violations, not just raise on the first error. Use pandas or Great Expectations.

Write an executive summary based on the following data findings. Context: [e.g. Q1 2026 performance review for a SaaS product] Key metrics: - [metric 1]: [value and change vs prior period] - [metric 2]: [value and change] - [metric 3]: [value and change] Notable events: [e.g. launched feature X in February, lost a major account in March] Write a 150-200 word executive summary. Lead with the most important insight, be direct about problems, and end with the top 2-3 priorities for next quarter.

Write a concise weekly metrics report for the following numbers. Team / product: [e.g. Growth team, mobile app] Week: [e.g. week of March 17, 2026] Metrics this week: - [metric]: [value] vs [last week value] - [metric]: [value] vs [last week value] Highlights: [any wins or notable events] Concerns: [any drops or issues] Format it as a short Slack or email update: 3-5 bullet points, plain language, no jargon. Flag items that need attention.

Write a clear summary of my A/B test results for a non-technical audience. Test name: [e.g. New checkout flow] Hypothesis: [e.g. Simplifying the checkout to 2 steps will increase conversion] Control vs variant: [describe the change] Results: - Primary metric: [e.g. conversion rate] — Control: [X%], Variant: [Y%], relative lift: [Z%] - Statistical significance: [e.g. p=0.03, 95% CI: +1.2% to +4.8%] - Secondary metrics: [any guardrail metrics and their results] Test duration: [e.g. 14 days, 25,000 users per group] Explain what the results mean, whether to ship the change, and any caveats.

Write a data dictionary for the following table or dataset. Table name: [e.g. orders] Columns: - [column_name]: [data type, e.g. INT] — [brief description of what it contains] - [column_name]: [data type] — [description] - [add more as needed] Audience: [e.g. new data analysts joining the team] For each column include: name, data type, description, example values, whether it can be null, and any known quirks or gotchas. Format it as a clean markdown table.