DATAGEN automates hypothesis generation, data analysis, and report writing. Operations teams use it to accelerate research and decision-making. It connects to Python workflows and supports Claude agents.
git clone https://github.com/starpig1129/DATAGEN.githttps://github.com/starpig1129/DATAGEN/docs/QUICKSTART.md
1. **Define Your Observation**: Clearly articulate the metric decline or anomaly you're investigating, including the time period and scope (e.g., '25% increase in cart abandonment in the US web app between March 10-15'). Use tools like Google Analytics, Mixpanel, or internal dashboards to confirm the baseline. 2. **Customize the Prompt**: Replace [OBSERVED_METRIC_DECLINE], [PRODUCT/SECTION/GEOGRAPHY], and [TIME_PERIOD] with your specific scenario. For advanced users, add constraints like 'Only consider hypotheses that can be tested with SQL queries' or 'Exclude hypotheses related to seasonal trends.' 3. **Run the Analysis**: Paste the prompt into your AI tool (Claude, ChatGPT, or a Claude agent) and review the generated hypotheses. Use the 'Expected Insight' column to shortlist the top 3-5 hypotheses for further validation. 4. **Validate with Data**: For each shortlisted hypothesis, use tools like: - **SQL queries** (BigQuery, Snowflake) to analyze funnel data or user segments. - **A/B testing platforms** (Optimizely, VWO) to test UI/UX changes. - **Support ticket analysis** (Zendesk, Intercom) to identify user-reported issues. - **Competitor monitoring tools** (SEMrush, SimilarWeb) to track external factors. 5. **Iterate and Refine**: If initial hypotheses are invalidated, regenerate the prompt with updated observations (e.g., 'After ruling out Hypothesis 1, focus on hypotheses that explain the remaining 12% decline'). Save successful hypotheses in a shared document for future reference.
Automate the generation of research hypotheses based on existing datasets.
Perform complex data analysis using machine learning techniques without manual coding.
Generate dynamic visualizations and reports from data files with a single command.
Conduct literature searches and integrate findings into research reports automatically.
No install command available. Check the GitHub repository for manual installation instructions.
git clone https://github.com/starpig1129/DATAGENCopy the install command above and run it in your terminal.
Launch Claude Code, Cursor, or your preferred AI coding agent.
Use the prompt template or examples below to test the skill.
Adapt the skill to your specific use case and workflow.
Generate 10 actionable hypotheses to explain [OBSERVED_METRIC_DECLINE] in [PRODUCT/SECTION/GEOGRAPHY] during [TIME_PERIOD]. For each hypothesis, include: (1) a clear statement, (2) the data source needed to validate it, and (3) a proposed analysis method. Format as a priority-ranked markdown table with columns: Rank | Hypothesis | Data Source | Analysis Method | Expected Insight. Prioritize hypotheses that can be tested with existing data within 2 weeks.
Here are 10 prioritized hypotheses to explain the 18% drop in 'Daily Active Users' for the Mobile App's 'Shopping Cart' feature between June 1-15, 2024, along with validation approaches:
| Rank | Hypothesis | Data Source | Analysis Method | Expected Insight |
|------|------------|-------------|-----------------|------------------|
| 1 | Checkout flow redesign (June 3) introduced a 3-step mandatory account creation that increased drop-off by 22% | App Analytics (Firebase), A/B Test Data | Compare funnel conversion rates between pre/post-redesign users | Identify specific steps causing friction |
| 2 | iOS 17.5 update (released June 5) broke the 'Save for Later' functionality, causing 15% of users to abandon carts | Crash Reports (Sentry), App Store Reviews | Correlate iOS version adoption with cart abandonment spikes | Pinpoint platform-specific issues |
| 3 | Competitor launched a 10% cashback promotion on June 8, diverting 8% of our users | Competitor Price Tracking Data, App Session Data | Analyze user traffic sources and session duration changes | Quantify competitive impact |
| 4 | Email campaign 'Summer Sale' sent on June 10 had a misleading subject line ('50% Off Everything'), leading to 12% higher bounce rates | Email Marketing Platform (Mailchimp), App Traffic Data | Compare email open/click rates with app engagement metrics | Assess campaign misalignment |
| 5 | Payment processor (Stripe) experienced outages on June 11-12, causing 7% of transactions to fail | Payment Gateway Logs, Error Tracking (Datadog) | Analyze transaction failure rates by time and payment method | Identify infrastructure vulnerabilities |
| 6 | New 'Guest Checkout' feature (released May 28) had a bug preventing order completion for 5% of users | Feature Flag Data, Support Tickets | Cross-reference support tickets with feature usage logs | Detect silent failures in new features |
| 7 | Summer heatwave reduced outdoor shopping activity, correlating with a 6% drop in mobile app usage | Weather Data API, App Session Data | Run regression analysis between temperature and DAU | Link external factors to usage patterns |
| 8 | 'Limited Time Offer' banner (added June 7) was placed below the fold, reducing visibility by 9% | UI Heatmap Data (Hotjar), Conversion Funnel Data | Compare banner interaction rates with pre-banner period | Optimize UI placement |
| 9 | Affiliate program commission rates were cut by 30% on June 1, reducing partner-driven traffic by 4% | Affiliate Tracking Platform, Referral Data | Analyze traffic sources and conversion rates by affiliate | Evaluate program ROI changes |
| 10 | Server-side caching issue on June 13 caused 11% slower page loads in the cart section | Performance Monitoring (New Relic), App Logs | Compare load times between cached/uncached requests | Identify backend performance bottlenecks |
**Next Steps:**
1. Prioritize hypotheses 1, 2, and 5 for immediate investigation (highest impact, fastest validation).
2. Schedule a meeting with the mobile team to review the checkout flow redesign (Hypothesis 1).
3. Set up a dashboard in Looker to monitor these metrics in real-time for early detection of similar issues.Your one-stop shop for church and ministry supplies.
Automate your browser workflows effortlessly
Create and collaborate on interactive animations with powerful, user-friendly tools.
IronCalc is a spreadsheet engine and ecosystem
Enterprise workflow automation and service management platform
Automate your spreadsheet tasks with AI power
Take a free 3-minute scan and get personalized AI skill recommendations.
Take free scan