Data Analysis Skill

Description

This skill helps analyze datasets and create meaningful visualizations. It can handle CSV files, perform statistical analysis, and generate various types of plots.

Usage Instructions

When a user requests data analysis:

1. •Load the dataset: Use pandas to read the data file
2. •Inspect the data: Check shape, columns, data types, and basic statistics
3. •Clean the data: Handle missing values and outliers if necessary
4. •Perform analysis: Calculate relevant statistics based on user's question
5. •Create visualizations: Generate appropriate plots (line, bar, scatter, etc.)
6. •Save results: Export results and visualizations

Available Resources

Scripts

• •

  scripts/analyze.py: Core analysis functions

  • •load_dataset(filepath): Load data from various formats
  • •basic_statistics(df): Calculate descriptive statistics
  • •detect_outliers(df, column): Identify outliers
  • •correlation_analysis(df): Compute correlations
• •

  scripts/visualize.py: Visualization utilities

  • •plot_distribution(df, column): Create distribution plots
  • •plot_correlation_matrix(df): Visualize correlation heatmap
  • •plot_time_series(df, date_col, value_col): Time series plots
  • •save_plot(fig, filename): Save figure to file

References

• •references/examples.md: Usage examples and common patterns
• •references/best_practices.md: Data analysis best practices

Example Prompts

• •"Analyze this CSV file and show me the trends"
• •"Create a visualization of the sales data by month"
• •"Find correlations in this dataset"
• •"Identify outliers in the price column"
• •"Generate a statistical summary of the data"

Output Format

Analysis results should include:

1. •Data overview (shape, columns, types)
2. •Statistical summary
3. •Key insights and findings
4. •Visualizations (saved as PNG files)
5. •Recommendations or next steps

Notes

• •Always inspect data before analysis
• •Handle missing values appropriately
• •Choose visualizations that match the data type
• •Provide clear explanations of findings
• •Save all outputs for user reference