LLM Data Narrative Generator

This project automates data analysis and narrative generation using a local Large Language Model (Qwen2.5 1.5B) downloaded from Kaggle.

Features

• Automated Data Analysis: Reads data from Excel files using openpyxl.
• Chart Generation: Creates charts (Line, Bar, Scatter) using matplotlib.
• Local LLM Inference: Uses a locally downloaded Qwen2.5 1.5B model via transformers to generate insightful narratives based on the data.
• Kaggle Integration: Automatically downloads models from Kaggle using kagglehub.

Prerequisites

• Python 3.8+
• Kaggle Account (for model download)

Setup

1. Clone the repository:

   git clone <repository_url>
   cd <repository_name>

2. Install dependencies:

   pip install -r requirements.txt

3. Kaggle Credentials: You need to authenticate with Kaggle to download the model. You can do this in one of the following ways:

   • Environment Variables: Set KAGGLE_USERNAME and KAGGLE_KEY.
   • Config File: Place your kaggle.json key in ~/.kaggle/kaggle.json.
   • Script Configuration: Edit llm_data_narrative3.py and set KAGGLE_USERNAME and KAGGLE_KEY variables at the top.

Usage

Run the main script:

python llm_data_narrative3.py

The script will:

1. Download the Qwen2.5 1.5B model from Kaggle (if not already present).
2. Create a sample Excel file (sample_data.xlsx) if one doesn't exist.
3. Analyze the data and generate a chart (analysis_chart.png).
4. Generate a text narrative (narrative.txt) describing the data and the chart.

Configuration

You can configure the script by editing the variables at the top of llm_data_narrative3.py:

• EXCEL_PATH: Path to your input Excel file.
• SHEET_NAME: Name of the sheet to analyze (defaults to active sheet).
• CHART_TYPE: Type of chart to generate ("line", "bar", "scatter").
• X_COL, Y_COL: Column indices for chart axes.

Output

• analysis_chart.png: The generated chart.
• narrative.txt: The AI-generated narrative explaining the data.
• llm_prompt.txt: The prompt sent to the LLM.