data-storyteller
について
データストーリーテラーは、CSV/Excelファイルを自動的にパターン検出し、平易な英語による洞察と視覚化を備えたナラティブレポートに変換します。pandasのDataFrameまたはファイルから直接、統計サマリーやPDFエクスポートを含む完全な分析ワークフローを提供します。このスキルを使用して、生の表形式データから迅速に、コード不要の説得力あるレポートを作成できます。
クイックインストール
Claude Code
推奨/plugin add https://github.com/majiayu000/claude-skill-registrygit clone https://github.com/majiayu000/claude-skill-registry.git ~/.claude/skills/data-storytellerこのコマンドをClaude Codeにコピー&ペーストしてスキルをインストールします
ドキュメント
Data Storyteller
Automatically transform raw data into compelling, insight-rich reports. Upload any CSV or Excel file and get back a complete analysis with visualizations, statistical summaries, and narrative explanations - all without writing code.
Core Workflow
1. Load and Analyze Data
from scripts.data_storyteller import DataStoryteller
# Initialize with your data file
storyteller = DataStoryteller("your_data.csv")
# Or from a pandas DataFrame
import pandas as pd
df = pd.read_csv("your_data.csv")
storyteller = DataStoryteller(df)
2. Generate Full Report
# Generate comprehensive report
report = storyteller.generate_report()
# Access components
print(report['summary']) # Executive summary
print(report['insights']) # Key findings
print(report['statistics']) # Statistical analysis
print(report['visualizations']) # Generated chart info
3. Export Options
# Export to PDF
storyteller.export_pdf("analysis_report.pdf")
# Export to HTML (interactive charts)
storyteller.export_html("analysis_report.html")
# Export charts only
storyteller.export_charts("charts/", format="png")
Quick Start Examples
Basic Analysis
from scripts.data_storyteller import DataStoryteller
# One-liner full analysis
DataStoryteller("sales_data.csv").generate_report().export_pdf("report.pdf")
Custom Analysis
storyteller = DataStoryteller("data.csv")
# Focus on specific columns
storyteller.analyze_columns(['revenue', 'customers', 'date'])
# Set analysis parameters
report = storyteller.generate_report(
include_correlations=True,
include_outliers=True,
include_trends=True,
time_column='date',
chart_style='business'
)
Features
Auto-Detection
- Column Types: Numeric, categorical, datetime, text, boolean
- Data Quality: Missing values, duplicates, outliers
- Relationships: Correlations, dependencies, groupings
- Time Series: Trends, seasonality, anomalies
Generated Visualizations
| Data Type | Charts Generated |
|---|---|
| Numeric | Histogram, box plot, trend line |
| Categorical | Bar chart, pie chart, frequency table |
| Time Series | Line chart, decomposition, forecast |
| Correlations | Heatmap, scatter matrix |
| Comparisons | Grouped bar, stacked area |
Narrative Insights
The storyteller generates plain-English insights including:
- Executive summary of key findings
- Notable patterns and anomalies
- Statistical significance notes
- Actionable recommendations
- Data quality warnings
Output Sections
1. Executive Summary
High-level overview of the dataset and key findings in 2-3 paragraphs.
2. Data Profile
- Row/column counts
- Memory usage
- Missing value analysis
- Duplicate detection
- Data type distribution
3. Statistical Analysis
For each numeric column:
- Central tendency (mean, median, mode)
- Dispersion (std dev, IQR, range)
- Distribution shape (skewness, kurtosis)
- Outlier count
4. Categorical Analysis
For each categorical column:
- Unique values count
- Top/bottom categories
- Frequency distribution
- Category balance assessment
5. Correlation Analysis
- Correlation matrix with significance
- Strongest relationships highlighted
- Multicollinearity warnings
6. Time-Based Analysis
If datetime column detected:
- Trend direction and strength
- Seasonality patterns
- Year-over-year comparisons
- Growth rate calculations
7. Visualizations
Auto-generated charts saved to report:
- Distribution plots
- Trend charts
- Comparison charts
- Correlation heatmaps
8. Recommendations
Data-driven suggestions:
- Columns needing attention
- Potential data quality fixes
- Analysis suggestions
- Business implications
Chart Styles
# Available styles
styles = ['business', 'scientific', 'minimal', 'dark', 'colorful']
storyteller.generate_report(chart_style='business')
Configuration
storyteller = DataStoryteller(df)
# Configure analysis
storyteller.config.update({
'max_categories': 20, # Max categories to show
'outlier_method': 'iqr', # 'iqr', 'zscore', 'isolation'
'correlation_threshold': 0.5,
'significance_level': 0.05,
'date_format': 'auto', # Or specify like '%Y-%m-%d'
'language': 'en', # Narrative language
})
Supported File Formats
| Format | Extension | Notes |
|---|---|---|
| CSV | .csv | Auto-detect delimiter |
| Excel | .xlsx, .xls | Multi-sheet support |
| JSON | .json | Records or columnar |
| Parquet | .parquet | For large datasets |
| TSV | .tsv | Tab-separated |
Example Output
Sample Executive Summary
"This dataset contains 10,847 records across 15 columns, covering sales transactions from January 2023 to December 2024. Revenue shows a strong upward trend (+23% YoY) with clear seasonal peaks in Q4. The top 3 product categories account for 67% of total revenue. Notable finding: Customer acquisition cost has increased 15% while retention rate dropped 8%, suggesting potential profitability concerns worth investigating."
Sample Insight
"Strong correlation detected between marketing_spend and new_customers (r=0.78, p<0.001). However, this relationship weakens significantly after $50K monthly spend, suggesting diminishing returns beyond this threshold."
Best Practices
- Clean data first: Remove obvious errors before analysis
- Name columns clearly: Helps auto-detection and narratives
- Include dates: Enables time-series analysis
- Provide context: Tell the storyteller what the data represents
Limitations
- Maximum recommended: 1M rows, 100 columns
- Complex nested data may need flattening
- Images/binary data not supported
- PDF export requires reportlab package
Dependencies
pandas>=2.0.0
numpy>=1.24.0
matplotlib>=3.7.0
seaborn>=0.12.0
scipy>=1.10.0
reportlab>=4.0.0
openpyxl>=3.1.0
GitHub リポジトリ
関連スキル
content-collections
メタThis skill provides a production-tested setup for Content Collections, a TypeScript-first tool that transforms Markdown/MDX files into type-safe data collections with Zod validation. Use it when building blogs, documentation sites, or content-heavy Vite + React applications to ensure type safety and automatic content validation. It covers everything from Vite plugin configuration and MDX compilation to deployment optimization and schema validation.
creating-opencode-plugins
メタThis skill provides the structure and API specifications for creating OpenCode plugins that hook into 25+ event types like commands, files, and LSP operations. It offers implementation patterns for JavaScript/TypeScript modules that intercept and extend the AI assistant's lifecycle. Use it when you need to build event-driven plugins for monitoring, custom handling, or extending OpenCode's capabilities.
evaluating-llms-harness
テストThis Claude Skill runs the lm-evaluation-harness to benchmark LLMs across 60+ standardized academic tasks like MMLU and GSM8K. It's designed for developers to compare model quality, track training progress, or report academic results. The tool supports various backends including HuggingFace and vLLM models.
sglang
メタSGLang is a high-performance LLM serving framework that specializes in fast, structured generation for JSON, regex, and agentic workflows using its RadixAttention prefix caching. It delivers significantly faster inference, especially for tasks with repeated prefixes, making it ideal for complex, structured outputs and multi-turn conversations. Choose SGLang over alternatives like vLLM when you need constrained decoding or are building applications with extensive prefix sharing.
