https://github.com/ascender1729/vodafone-financial-analysis
Automated financial table extraction and standardization from Vodafone's annual report using GPT-4o-mini
https://github.com/ascender1729/vodafone-financial-analysis
automation balance-sheet crediflow-ai csv financial-analysis financial-tables gpt-4o-mini machine-learning ocr openai pandas pdf-extraction pdf2image pypdf2 pytesseract standardization striprtf vodafone
Last synced: about 2 months ago
JSON representation
Automated financial table extraction and standardization from Vodafone's annual report using GPT-4o-mini
- Host: GitHub
- URL: https://github.com/ascender1729/vodafone-financial-analysis
- Owner: ascender1729
- License: mit
- Created: 2025-02-22T22:16:21.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-02-22T22:47:33.000Z (over 1 year ago)
- Last Synced: 2025-07-03T13:57:25.378Z (about 1 year ago)
- Topics: automation, balance-sheet, crediflow-ai, csv, financial-analysis, financial-tables, gpt-4o-mini, machine-learning, ocr, openai, pandas, pdf-extraction, pdf2image, pypdf2, pytesseract, standardization, striprtf, vodafone
- Language: Rich Text Format
- Homepage: https://github.com/ascender1729/vodafone-financial-analysis
- Size: 797 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# vodafone-financial-analysis
> Automated financial table extraction and standardization from Vodafone's annual report using GPT-4o-mini
## Table of Contents
- [Overview](#overview)
- [Project Structure](#project-structure)
- [Features](#features)
- [Solution Architecture](#solution-architecture)
- [Usage](#usage)
- [Implementation Details](#implementation-details)
- [Results](#results)
- [Future Enhancements](#future-enhancements)
- [License](#license)
## Overview
This project automates the extraction and standardization of financial tables from Vodafone's annual report using GPT-4o-mini. Developed for the Crediflow AI Assignment, it addresses:
1. Extraction of financial tables from PDF report
2. Standardization of Balance Sheet terminology
3. Future optimization proposals
## Project Structure
```
vodafone-financial-analysis/
├── extracted_data/
│ ├── extracted_text.txt # Raw extracted text
│ ├── extraction_validation.log # Validation logs
│ └── processed_vocabulary.json # Processed vocabulary terms
├── financial_tables/
│ ├── INCOME_STATEMENT.csv
│ ├── STATEMENT_OF_COMPREHENSIVE_INCOME.csv
│ ├── STATEMENT_OF_FINANCIAL_POSITION.csv
│ └── financial_tables.md
├── standardized_output/
│ ├── standardized_balance_sheet.csv
│ └── standardized_balance_sheet.md
├── input/
│ ├── vocabulary.rtf # Vocabulary list
│ └── vodafone_annual_report.pdf # Source PDF
├── docs/
│ ├── workflow.svg # Solution workflow diagram
│ └── Crediflow_AI_Assignment.ipynb
└── README.md
```
## Features
- PDF text extraction with OCR fallback
- GPT-4o-mini powered table identification
- Financial term standardization
- Multiple output formats (Markdown, CSV)
- Detailed validation and logging
## Solution Architecture
[//]: # "Insert workflow diagram here"
## Usage
[](https://colab.research.google.com/drive/1hrjoKFIDsxRZfuGJYMr_vTO_YlUR3wXd?usp=sharing)
### Quick Start
1. Add input files from `/input` directory to the notebook
2. Run text extraction notebook
3. Process financial tables
4. Check outputs in respective directories
## Implementation Details
### Question 1: Table Extraction
Successfully extracted three financial tables:
| Table Type | Status | Output Location |
|------------|--------|-----------------|
| Income Statement | ✅ | financial_tables/INCOME_STATEMENT.csv |
| Balance Sheet | ✅ | financial_tables/STATEMENT_OF_FINANCIAL_POSITION.csv |
| Comprehensive Income | ✅ | financial_tables/STATEMENT_OF_COMPREHENSIVE_INCOME.csv |
### Question 2: Balance Sheet Standardization
Implemented term standardization using:
- GPT-4o-mini for semantic matching
- Predefined vocabulary from RTF file
- Structure preservation
- Numerical accuracy validation
Results available in:
- standardized_output/standardized_balance_sheet.csv
- standardized_output/standardized_balance_sheet.md
### Question 3: Future Optimizations
#### Technical Improvements
- Batch processing capabilities
- Enhanced error handling
- Improved OCR accuracy
#### Feature Additions
- Support for multiple file formats
#### Quality Enhancements
- Comprehensive validation
- Performance optimization
- Enhanced logging
- Better error reporting
## Results
Successfully processed Vodafone's annual report with:
- Three financial tables extracted
- Standardized Balance Sheet terminology
- Structured outputs in CSV and MD formats
- Preserved numerical accuracy
- Maintained hierarchical relationships
## Dependencies
- Python 3.8+
- OpenAI API access
- Required libraries:
```
pytesseract
pdf2image
PyPDF2
pandas
openai
striprtf
```
## License
MIT License
```
Created as part of the Crediflow AI Assignment
```