Policy Text Analysis using NLP

📌 Project Overview

This project applies NLP-based text analysis to sample insurance policy wording data in order to compare old and new policy versions, identify wording differences, detect benefit-level changes, and analyze coverage amount changes.

The project uses Python, Pandas, Matplotlib, and Excel for text preprocessing, policy comparison, change detection, visualization, and report generation.

Large Language Models (LLMs), such as ChatGPT, were used manually as an AI-assisted tool to summarize key wording differences and prepare clear reporting notes.

This project uses sample training data created for analysis purposes and does not contain real customer or confidential policy information.

---

🎯 Objectives

• Clean and prepare sample policy wording data
• Apply basic NLP text preprocessing
• Compare old and new versions of policy text
• Identify added and removed wording
• Detect benefit-level and coverage amount changes
• Classify changes as Major, Moderate, or Minor
• Use LLM-assisted summaries to support reporting
• Create simple visualizations for analysis
• Export structured findings into an Excel report

---

🗂️ Dataset

The dataset contains sample insurance policy records.

Column	Description
Policy_ID	Unique policy record ID
Policy_Group	Group used to compare old and new policy versions
Version	Old or New policy version
Category	Policy category such as Medical, Dental, Vision, or Life Insurance
Policy_Text	Policy wording text
Benefit_Level	Benefit level such as Low, Medium, or High
Coverage_Amount	Coverage amount value

Each policy group includes an old and new version to allow text comparison and change detection.

---

🛠️ Tools Used

Tool	Purpose
Python	Data analysis and text processing
Pandas	Data cleaning and comparison
Matplotlib	Data visualization
Excel	Final report output
Visual Studio Code	Development environment
LLMs	AI-assisted summarization and reporting notes

---

📁 Project Structure

Policy-Text-Analysis-using-NLP/
│
├── data/
│   └── policy_samples.csv
│
├── notebooks/
│   └── policy_text_analysis.ipynb
│
├── outputs/
│   └── policy_comparison_report.xlsx
│
├── images/
│   ├── Policy_Report.png
│   └── Coverage_Visualization.png
│
├── README.md
└── requirements.txt

---

🔍 Analysis Workflow

1. Created and loaded sample insurance policy wording data
2. Cleaned and standardized policy text
3. Applied basic NLP text preprocessing
4. Compared old and new policy versions by policy group
5. Identified added and removed words
6. Detected benefit-level and coverage amount changes
7. Classified policy changes by impact level
8. Used LLM-assisted summaries for reporting notes
9. Created visualizations to show policy changes
10. Exported the final analysis into an Excel report

---

🤖 NLP and LLM-Assisted Analysis

This project combines basic NLP-style text processing with AI-assisted summarization.

The NLP-related analysis includes:

• Text cleaning and standardization
• Old vs. new policy text comparison
• Added and removed word detection
• Change detection across policy versions
• Benefit-level and coverage change analysis
• Structured text findings for reporting

Large Language Models (LLMs), such as ChatGPT, were used manually as an AI-assisted tool to summarize policy wording differences and generate reporting notes.

Prompt used:

Compare the following two policy wording samples.

Old Policy:
[Old policy text]

New Policy:
[New policy text]

Please summarize:
1. Key wording differences
2. Added benefits
3. Removed or changed terms
4. Benefit-level impact
5. A short reporting note for a data analyst

---

📊 Visualizations

The notebook includes visualizations for:

• Policy changes by change type
• Coverage amount change by policy group
• Average coverage change by category

These visualizations help summarize the impact of wording and benefit-level changes across different policy categories.

---

📸 Project Screenshots

Excel Policy Report

Coverage Amount Change Visualization

---

📤 Output

The final output is an Excel report:

outputs/policy_comparison_report.xlsx

The report includes:

Output Field	Description
LLM_Assisted_Summary	AI-assisted summary of the policy change
Policy_Group	Policy comparison group
Category	Policy category
Old_Benefit_Level	Previous benefit level
New_Benefit_Level	Updated benefit level
Coverage_Change	Difference in coverage amount
Added_Words	New wording added
Removed_Words	Wording removed from old policy
Change_Type	Major, Moderate, or Minor change

---

✅ Key Results

Metric	Value
Total policies analyzed	10
Total policy groups compared	5
Major changes identified	5
Average coverage change	1,210

---

▶️ How to Run the Project

1. Clone or download this repository.

2. Install the required libraries:

pip install -r requirements.txt

3. Open the analysis notebook in Visual Studio Code:

notebooks/policy_text_analysis.ipynb

4. Select a Python environment.

5. Run all cells in order.

6. Review the generated Excel report in the outputs folder.

---

💡 Project Value

This project demonstrates practical data analysis and text analysis skills, including:

• Data cleaning
• NLP-based text comparison
• Text preprocessing
• Change detection
• Policy comparison
• Benefit-level analysis
• Reporting
• Visualization
• AI-assisted summarization

It is especially relevant for Data Analyst and entry-level NLP / AI-assisted analysis roles involving document comparison, text analysis, reporting, and the use of Large Language Models as productivity tools.