Data Analysis & Visualization

Help me build an interactive dashboard to analyze user retention data:

Data sources:
- User registration data (users.csv)
- User activity log (activity.csv)

Functional requirements:
- Display 7-day, 30-day, and 90-day retention rate curves
- Group comparison by user source (advertising, natural traffic,
  recommendation)
- Time range filter
- Bar charts showing weekly active user trends

Technology stack: Use React + Recharts to generate a web page
that can be run directly.

Help me analyze this sales data (sales_2024.csv):

1. Give me an overview of the data (number of rows, columns,
   missing values)
2. Generate descriptive statistics (mean, median, standard deviation)
3. Identify outliers
4. Do correlation analysis to see which factors affect sales
5. Draw distribution charts and trend charts of key indicators
6. Summarize 3-5 key findings

Use this customer churn data (churn_data.csv) to train a
prediction model:

1. Data preprocessing (handle missing values, standardize
   numerical features)
2. Feature engineering (generate useful new features)
3. Train several models (logistic regression, random forest, XGBoost)
4. Compare model performance (precision, recall, F1, AUC)
5. Generate feature importance analysis
6. Give me the prediction code of the best model that can
   predict new customers

Help me set up automatic monitoring:

- Check this BigQuery data table every morning at 9am
- Send an alert if daily active users are 20% lower than
  the 7-day average
- Send an alert if the error rate exceeds 5%
- Generate daily data summary and send to my email

This approach seems too complex. Try something simpler
with fewer dependencies.

Process this data with Python and visualize with JavaScript
using D3.js for interactive charts.

Perform customer segmentation analysis on this data:

[Upload customer_data.csv]

Data includes:
- Demographics (age, location, income)
- Purchase history (frequency, value, recency)
- Engagement metrics (email opens, website visits)

Tasks:
1. Perform RFM analysis (Recency, Frequency, Monetary)
2. Use K-means clustering to identify 4-5 customer segments
3. Profile each segment (characteristics, behaviors)
4. Visualize segments with scatter plots and radar charts
5. Recommend marketing strategies for each segment
6. Export segment assignments for use in CRM

Create a sales forecasting model:

Data: monthly_sales.csv (3 years of historical data)

Requirements:
1. Decompose time series (trend, seasonality, residuals)
2. Check for stationarity (ADF test)
3. Train multiple models:
   - ARIMA
   - Prophet (Facebook's forecasting tool)
   - LSTM (if patterns are complex)
4. Compare model performance (RMSE, MAE, MAPE)
5. Forecast next 6 months
6. Create confidence intervals
7. Visualize historical data + forecasts with interactive chart

Explain which model performed best and why.

Analyze results from our A/B test:

[Upload ab_test_results.csv]

Test details:
- Control: Current checkout flow
- Variant: New one-click checkout
- Metrics: Conversion rate, average order value, completion time
- Sample size: 10,000 users per variant

Analysis needed:
1. Calculate statistical significance (p-value, confidence intervals)
2. Check for sample ratio mismatch
3. Analyze by user segments (new vs. returning, device type)
4. Calculate practical significance (effect size)
5. Estimate revenue impact if we roll out variant
6. Visualize results with clear charts
7. Provide go/no-go recommendation

Be conservative with statistical interpretation.

Build a cohort retention analysis:

Data: user_activity.csv with user_id, signup_date, activity_date

Analysis:
1. Group users by signup month (cohorts)
2. Calculate retention for each cohort at:
   - Day 1, 7, 14, 30, 60, 90
3. Create cohort retention heatmap
4. Identify which cohorts have best retention
5. Analyze if there are seasonal patterns
6. Compare cohorts before/after a major feature launch (June 2024)
7. Build interactive dashboard to explore different cohort groupings

Help me understand what's driving retention differences.

Help me optimize this slow SQL query:

[Paste SQL query]

Database: PostgreSQL
Table size: 50M rows
Current execution time: 45 seconds

Please:
1. Explain what the query is doing
2. Identify performance bottlenecks
3. Suggest optimizations (indexes, query rewrite, etc.)
4. Explain the execution plan
5. Provide the optimized version
6. Estimate expected performance improvement

Also suggest what indexes I should create.

Create an automated weekly analytics report:

Data sources:
- PostgreSQL database (user events)
- Google Analytics (via API)
- Stripe (via API for revenue data)

Report sections:
1. Executive Summary
   - Key metrics vs. previous week
   - Notable changes (flag >10% changes)

2. User Growth
   - New signups (daily trend)
   - Activation rate
   - Growth rate by channel

3. Engagement
   - DAU/WAU/MAU trends
   - Feature usage breakdown
   - Session duration analysis

4. Revenue
   - MRR and growth rate
   - New vs. expansion vs. churn
   - Customer LTV by cohort

Output:
- HTML report with embedded charts
- PDF version for distribution
- Send via email every Monday 8am

Automate this to run weekly. (Pro/Max plan)

Set up data quality checks:

Dataset: user_events table (BigQuery)

Checks to implement:
1. Completeness
   - No null values in critical fields
   - Expected row count (±20% of 7-day average)

2. Uniqueness
   - No duplicate event_ids
   - User_id format validation

3. Timeliness
   - Events processed within 1 hour
   - No data gaps > 30 minutes

4. Validity
   - Timestamps in reasonable range
   - Numeric fields within expected bounds
   - Category values match allowed list

Alert me if any check fails. Run checks every hour.
Provide dashboard showing data quality trends.

Help me build a feature store for ML models:

Raw data: user_profiles.csv, transactions.csv, events.csv

Features to engineer:
1. User features:
   - Total transactions (all time)
   - Average transaction value
   - Days since last transaction
   - Transaction frequency (per month)
   - Preferred categories

2. Behavioral features:
   - Page views last 7/30 days
   - Session count last 7/30 days
   - Engagement score (custom calculation)

3. Temporal features:
   - Day of week patterns
   - Time of day patterns

Requirements:
- Update features daily
- Store in format ready for model training
- Handle missing values appropriately
- Include feature documentation
- Version control for features

Create pipeline that computes and updates these features.

Create an interactive dashboard where users can:
- Filter by date range with a slider
- Toggle between different metrics
- Hover to see detailed values
- Click on segments to drill down
- Export current view as PNG

Use Plotly or Recharts for interactivity.

I have a 10GB CSV file that's too large to process in memory.

Help me:
1. Process it in chunks
2. Perform aggregations efficiently
3. Create summary statistics
4. Sample the data for visualization
5. Identify outliers without loading everything

Use appropriate tools (dask, polars, or chunking strategies).

This dashboard query is taking 30+ seconds to load.

Current query: [paste SQL]

Please optimize by:
1. Identifying unnecessary joins
2. Suggesting appropriate indexes
3. Rewriting with better structure
4. Using materialized views if appropriate
5. Implementing caching strategy

Goal: <5 second load time