csv-data-visualizer

CSV Data Visualizer

Overview

This skill enables comprehensive data visualization and analysis for CSV files. It provides three main capabilities: (1) creating individual interactive visualizations using Plotly, (2) automatic data profiling with statistical summaries, and (3) generating multi-plot dashboards. The skill is optimized for exploratory data analysis, statistical reporting, and creating presentation-ready visualizations.

When to Use This Skill

Invoke this skill when users request:

• "Visualize this CSV data"
• "Create a histogram/scatter plot/box plot from this data"
• "Show me the distribution of [column]"
• "Generate a dashboard for this dataset"
• "Profile this CSV file" or "Analyze this data"
• "Create a correlation heatmap"
• "Show trends over time"
• "Compare [variable] across [categories]"

Core Capabilities

1. Individual Visualizations

Create specific chart types for detailed analysis using the visualize_csv.py script.

Available Chart Types:

Statistical Plots:

# Histogram - distribution of numeric data
python3 scripts/visualize_csv.py data.csv --histogram column_name --bins 30
 
# Box plot - show quartiles and outliers
python3

Relationship Analysis:

Time Series:

Categorical Data:

Output Formats: Specify output file with desired format extension:

2. Automatic Data Profiling

Generate comprehensive data quality and statistical reports using the data_profile.py script.

Text Report (default):

HTML Report:

JSON Report:

What the Profiler Provides:

• File information (size, dimensions)
• Dataset overview (shape, memory usage, duplicates)
• Column-by-column analysis (types, missing data, unique values)
• Missing data patterns and completeness
• Statistical summary for numeric columns (mean, std, quartiles, skewness, kurtosis)
• Categorical column analysis (frequency counts, most/least common values)
• Data quality checks (high missing data, duplicate rows, constant columns, high cardinality)

When to Use Profiling: Always recommend running data profiling BEFORE creating visualizations when:

• User is unfamiliar with the dataset
• Data quality is unknown
• Need to identify appropriate visualization types
• Exploring a new dataset for the first time

3. Multi-Plot Dashboards

Create comprehensive dashboards with multiple visualizations using the create_dashboard.py script.

Automatic Dashboard: Analyzes data types and automatically creates appropriate visualizations:

Custom output location:

Control number of plots:

Custom Dashboard from Config: Create a JSON configuration file specifying exact plots:

Dashboard Config Format:

Dashboard Plot Types:

• histogram: Distribution of numeric column
• box: Box plot, optionally grouped by category
• scatter: Relationship between two numeric columns
• bar: Count of categorical values
• correlation: Heatmap of numeric correlations

Workflow Decision Tree

Use this decision tree to determine the appropriate approach:

Best Practices

Starting Analysis

1. Always profile first for unfamiliar datasets: python3 scripts/data_profile.py data.csv
2. Review the profiling output to understand:
   • Column data types and ranges
   • Missing data patterns
   • Data quality issues
   • Statistical distributions

Choosing Visualizations

Consult references/visualization_guide.md for detailed guidance. Quick reference:

• Distribution: Histogram, box plot, violin plot
• Relationship: Scatter plot, correlation heatmap
• Time series: Line chart
• Categories: Bar chart (preferred) or pie chart (use sparingly)
• Comparison: Box plot grouped by category

Creating Dashboards

• Automatic dashboard: Good for initial exploration
• Custom dashboard: Better for presentations or specific analysis goals
• Limit plots: Keep to 6-9 plots maximum for readability
• Logical grouping: Group related visualizations together

Output Considerations

• HTML: Best for interactive exploration (zoom, pan, hover tooltips)
• PNG/PDF: Best for reports and presentations
• SVG: Best for publications requiring vector graphics

Dependencies

The scripts require these Python packages:

For static image export (PNG, PDF, SVG), also install:

Example Workflows

Exploratory Data Analysis

Report Generation

Interactive Dashboard

Troubleshooting

"Column not found" errors:

• Run data profiling to see exact column names
• CSV columns are case-sensitive
• Check for leading/trailing spaces in column names

Empty or incorrect visualizations:

• Verify data types (numeric vs categorical)
• Check for missing data in plotted columns
• Ensure sufficient non-null values exist

Script execution errors:

• Verify dependencies are installed: pip list | grep plotly
• Check Python version: Python 3.6+ required
• For image export issues, install kaleido: pip install kaleido

Resources

scripts/

• visualize_csv.py: Main visualization script with all chart types
• data_profile.py: Automatic data profiling and quality analysis
• create_dashboard.py: Multi-plot dashboard generator

references/

• visualization_guide.md: Comprehensive guide for choosing appropriate chart types, best practices, and common patterns