---
title: KPI Score Correlation Analysis
emoji: 📊
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: "5.33.0"
app_file: kpi_correlation_app.py
pinned: false
---

# KPI Score Correlation Analysis

This directory contains tools for analyzing correlations between IPM scores and axiia scores.

## Architecture

The code has been refactored to share common functionality:

- **`correlation_analysis_core.py`**: Core analysis module with shared functions
- **`csv_utils.py`**: Utilities for loading CSV/Excel files
- **`analyze_correlations_v2.py`**: Command-line interface (CLI)
- **`kpi_correlation_app.py`**: Gradio web interface

## Installation

Ensure you have the required dependencies:

```bash
pip install pandas numpy scipy matplotlib seaborn gradio pyyaml openpyxl xlrd
```

## Usage

### Command-Line Interface

The CLI tool is best for batch processing and automation:

```bash
# Basic usage
python3 analyze_correlations_v2.py -k kpi_file.csv -s scores_file.csv

# With custom output file
python3 analyze_correlations_v2.py -k kpi_file.csv -s scores_file.csv -o results.yaml

# With plots
python3 analyze_correlations_v2.py -k kpi_file.csv -s scores_file.csv -p

# Full example
python3 analyze_correlations_v2.py \
  -k ../../data/lenovo_kpi.csv \
  -s ../../data/lenovo-scores-0603.csv \
  -o score_corr.yaml \
  -p
```

Options:
- `-k, --kpi`: Path to KPI file (CSV or Excel)
- `-s, --scores`: Path to scores file (CSV)
- `-o, --output`: Output YAML file (default: score_corr.yaml)
- `-p, --plot`: Generate correlation plots

### Gradio Web Interface

The Gradio app provides an interactive UI with parameterized analysis:

```bash
# Basic usage (uses default scores file in same directory)
python3 kpi_correlation_app.py

# With custom scores file
python3 kpi_correlation_app.py --scores-file path/to/scores.csv

# Share publicly
python3 kpi_correlation_app.py --share

# Custom port
python3 kpi_correlation_app.py --port 8080

# Full example
python3 kpi_correlation_app.py \
  --scores-file ../../data/lenovo-scores-0603.csv \
  --port 7860
```

Options:
- `--scores-file`: Path to scores CSV file (default: lenovo-scores-0603.csv)
- `--share`: Create a public link
- `--port`: Port to run on (default: 7860)

The Gradio interface provides the following parameterized features:

1. **Data Selection**:
   - Choose between different KPI files (CSV/Excel)
   - Select specific score columns for analysis
   - Filter data by manager status and other criteria

2. **Analysis Parameters**:
   - Correlation method selection (Pearson/Spearman)
   - Confidence level adjustment
   - Sample size requirements
   - Outlier detection thresholds

3. **Visualization Options**:
   - Plot type selection (scatter, regression, etc.)
   - Color scheme customization
   - Figure size and DPI settings
   - Trend line display options

4. **Output Configuration**:
   - Export format selection (YAML/CSV/Excel)
   - Custom output file naming
   - Detailed vs. summary report options
   - Plot export settings

All parameters can be adjusted in real-time through the web interface, with immediate updates to the analysis results and visualizations.

## Input File Requirements

### KPI File
- Must contain an email column (case-insensitive)
- Must contain IPM columns for FY23/24 and FY24/25
- Supports CSV and Excel formats

### Scores File
- Must contain columns: `email`, `problem_score`, `ability_score`
- CSV format

## Output

### CLI Output
- Console output with data quality report and correlation analysis
- YAML file with detailed results
- Optional PNG plots (individual and combined)

### Gradio Output
- Interactive web interface
- Real-time analysis results
- Interactive scatter plots with trend lines
- Data quality statistics

## Correlation Pairs Analyzed

- **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

Each pair shows:
- Pearson correlation coefficient
- Spearman correlation coefficient
- Number of valid samples
- Data quality metrics

# HF deployment

git subtree push --prefix=data-analysis/kpi_score_analysis hfspace main