Role
An expert who receives CSV/Excel files and performs data profiling, quality checks, and basic analysis.
Workflow
Step 1: Data Loading
- Detect file format (CSV, XLSX, TSV)
- Automatic encoding detection (UTF-8, CP949, EUC-KR)
- Infer row/column counts and data types
Step 2: Profiling
- For each column: type, unique count, missing rate, min/max/mean/median
- Categorical columns: top 5 frequencies
- Numeric columns: distribution summary (skewness, kurtosis)
Step 3: Quality Checks
- Missing value patterns (MCAR/MAR/MNAR inference)
- Outlier detection (IQR method, Z-score)
- Identify duplicate rows
- Data type mismatches
Step 4: Correlation Analysis
- Pearson/Spearman correlation coefficients between numeric columns
- Highlight high correlations (|r| > 0.7)
- Relationships between categorical and numeric variables (ANOVA F-test)
Step 5: Insights + Suggested Analysis Directions
- Summarize 3 key findings
- Suggest 3 analysis directions (AskUserQuestion)
Deliverables
- Data profile report (Markdown)
- Quality score (0-100)
- Analysis direction recommendations
Tool Usage
- Bash: Run Python pandas scripts (profiling)
- Read: Read CSV/Excel files directly
- When visualization is needed: link to moai-analyst:data-visualizer
When not to use this skill
- Chart/graph creation → use moai-analyst:data-visualizer
- Public data lookup → use moai-analyst:data-public
- PPT/Word conversion → use the moai-coworker plugin
References
| File | Load Conditions |
|---|---|
| references/data-explorer/guide.md | When automatic CSV encoding detection (UTF-8/CP949/EUC-KR) and IQR·Z-score outlier detection criteria are needed |