docs/gradio_app_usage.md

KPI Correlation Analysis Gradio App Usage Guide

Overview

This Gradio app provides a simple web interface for analyzing correlations between IPM scores and axiia scores (problem_score and ability_score).

How to Run the App

1. Navigate to the app directory:

   cd data-analysis/kpi_score_analysis
   

2. Run the app:

   python3 kpi_correlation_app.py
   

3. Access the app:

   • The app will start and display a URL (typically http://127.0.0.1:7860)
   • Open this URL in your web browser

Using the App

1. Upload KPI File:

   • Click on the "Upload KPI File" area
   • Select your KPI file (supports .csv, .xls, .xlsx formats)
   • The file should contain:
     • Email column
     • FY23/24 IPM column
     • FY24/25 IPM column

2. Analyze:

   • Click the "Analyze Correlations" button
   • The app will process the data and match emails with the scores file

3. View Results:

   • Analysis Results: Text output showing:

     • Data quality report (number of records, matched emails)
     • Correlation analysis for each pair (AC, AD, BC, BD)
     • Pearson and Spearman correlation coefficients
     • P-values and sample sizes

   • Correlation Plots: Four scatter plots showing:

     • AC: Problem Score vs FY23/24 IPM
     • AD: Problem Score vs FY24/25 IPM
     • BC: Ability Score vs FY23/24 IPM
     • BD: Ability Score vs FY24/25 IPM

Example Files

You can test the app with these files:

• test_kpi.csv (in the same directory)
• ../../data/lenovo_kpi.csv
• ../../data/Copy of 联想 kpi copy.xlsx

Technical Details

• The app uses a fixed scores file (lenovo-scores-0603.csv) that must be in the same directory
• Handles both CSV and Excel files for KPI data
• Automatically converts percentage strings to numeric values
• Only analyzes records with matching emails and complete data
• Shows trend lines and correlation coefficients on plots

Troubleshooting

• "Error: scores file not found": Ensure lenovo-scores-0603.csv is in the same directory as the app
• "No matching emails found": Check that email addresses in KPI file match those in scores file
• Empty plots: This indicates insufficient data points for that correlation pair

Requirements

• Python 3.x
• Required packages: gradio, pandas, numpy, scipy, matplotlib, seaborn
• All dependencies should already be installed in the lenovo-reports environment