Data Science

Help me design a dashboard for:

**Audience:** [who will use this]
**Purpose:** [monitoring/analysis/reporting]
**Key Questions:** [what decisions will it inform]
**Data Sources:** [where data comes from]
**Update Frequency:** [real-time/daily/weekly]

Provide:

**1. KPI Selection**
- Primary metrics (3-5 max)
- Supporting metrics
- Metric definitions and calculations

**2. Layout Recommendation**
- Information hierarchy
- Section organization
- Visual flow

**3. Chart Types**
For each metric:
- Recommended visualization
- Why this chart type
- Interaction patterns

**4. Filtering & Drill-down**
- Global filters needed
- Drill-down paths
- Comparison options

**5. Alerts & Thresholds**
- What to highlight
- Threshold values
- Alert conditions

**6. Anti-patterns to Avoid**
- Common dashboard mistakes
- Specific to this use case

You are a senior data analyst. Create an analysis report for:

Dataset: [DESCRIBE YOUR DATA]
Business Question: [WHAT ARE YOU TRYING TO ANSWER]
Audience: [WHO WILL READ THIS]

Structure:
1. Executive Summary: Key findings in 3 bullets
2. Methodology: How you approached the analysis
3. Key Metrics: Define each metric, current values, trends
4. Insights: What the data tells us
5. Recommendations: Data-driven action items
6. Next Steps: Further analysis needed

Guide me through an EDA for this dataset:

**Dataset Description:** [what the data represents]
**Columns:** [list main columns and types]
**Business Question:** [what you're trying to learn]
**Tools:** [Python/R/SQL]

Provide code and explanations for:

**1. Data Overview**
- Shape, types, memory usage
- First/last rows inspection

**2. Data Quality**
- Missing values analysis
- Duplicate detection
- Outlier identification

**3. Univariate Analysis**
- Distributions of key variables
- Summary statistics
- Visualizations (histograms, box plots)

**4. Bivariate Analysis**
- Correlations
- Cross-tabulations
- Scatter plots for relationships

**5. Key Insights**
- Top 5 findings with implications
- Hypotheses to test
- Data quality issues to address

**6. Next Steps**
- Recommended analyses
- Features to engineer
- Questions for stakeholders

Tester Prompts

Help me choose the right ML model for my problem:

**Problem Type:** [classification/regression/clustering/etc.]
**Target Variable:** [what you're predicting]
**Features:** [types of input data]
**Dataset Size:** [rows and columns]
**Constraints:** [interpretability, latency, compute]
**Current Baseline:** [if any]

Provide:

**1. Problem Framing**
- Confirm problem type
- Success metrics to use
- Evaluation strategy

**2. Model Candidates**
For each recommended model:
- Why it fits this problem
- Pros and cons
- Hyperparameters to tune
- Computational requirements

**3. Recommended Approach**
- Start with: [simple baseline]
- Then try: [more complex options]
- Consider: [advanced techniques if needed]

**4. Implementation Checklist**
- Data preprocessing steps
- Feature engineering ideas
- Cross-validation strategy
- Overfitting prevention

**5. Code Skeleton**
- Basic implementation in [Python/sklearn/etc.]

You are a SQL expert. Write optimized SQL queries for:

Database: [POSTGRESQL/MYSQL/SQLITE/etc.]
Task: [DESCRIBE WHAT DATA YOU NEED]
Tables: [LIST RELEVANT TABLES AND KEY COLUMNS]
Constraints: [ANY FILTERS, DATE RANGES, LIMITS]

Provide:
1. The SQL query with comments explaining each part
2. Explanation of the approach
3. Index recommendations for performance
4. Alternative approaches if applicable

You are an expert Data Science Consultant. When a user describes a business problem:

1. **Clarify** (if needed): Ask 1-2 quick questions about their data and goals
2. **Classify**: Identify the problem type (prediction, classification, clustering, recommendation, optimization, causal analysis)
3. **Recommend**: Suggest 2-3 techniques ranked by complexity:
   - **Simple baseline**: Fast to implement, easy to explain
   - **Recommended approach**: Best balance of performance and effort
   - **Advanced option**: If they have time/resources

For each technique, briefly explain:
- Why it fits their problem
- What data they need
- Key pitfalls to avoid

Be direct and practical. Use their business language, not just ML jargon. Focus on what will actually work, not what's theoretically ideal.