Demo

View a Demo: https://vimeo.com/940595873?share=copy

Table of Contents 🗓

1. About The Project
2. Objectives and Strategies
3. Libraries Used
4. Getting Started
   • Prerequisites
   • Installation
5. Roadmap
6. Contact
7. Demo

About The Project 🚀

The objective of this project is to analyze and understand the Consumer Financial Protection Bureau (CFPB) consumer complaint narrative dataset, which consists of approximately 2.3 million observations with 18 columns. These columns include geographical, institutional, and other meta-information that can be leveraged to generate insights across a wide range of parameters.

(back to top)

Objectives and Strategies

1. Organizational Feedback

• Root Cause Analysis: Identify the departments or individuals responsible for complaints and generate feedback to the correct departments to minimize complaints and maximize customer satisfaction.
• Correlation between Cause and Sentiment: Determine the root cause of complaints and evaluate the associated sentiment to understand the intensity and sensitivity of the issues, which will help prioritize necessary changes.

2. Topic Modelling to Identify Emerging Issues

• Apply topic modelling to uncover common themes within the narratives and observe how these issues evolve, pinpointing newly emerging consumer concerns.

3. Text Complexity and Readability Analysis

• Evaluate the complexity and readability of the narratives to explore any correlation between the nature of a complaint and its resolution outcome.

4. Geographic Analysis of Complaints

• Regional Sentiment Analysis: Perform sentiment analysis segmented by state or ZIP code to discover geographic patterns in financial service experiences.
• Topic Modelling by Geography: Apply topic modelling on a per-region basis to identify prevalent issues in specific areas, providing insights into regional financial challenges.

5. Predictive Modelling for Resolution Outcomes

• Utilize the narratives and metadata (e.g., product type, submission format) to develop predictive models that forecast complaint resolution outcomes, identifying key factors that influence consumer satisfaction.

6. Benchmarking and Comparative Analysis

• Company Performance Analysis: Analyze complaints by company to benchmark their performance in handling these issues, comparing metrics such as average sentiment score, resolution time, and dispute rates.
• Product Type Analysis: Benchmark financial products to identify those associated with higher levels of dissatisfaction.

(back to top)

Getting Started

Prerequisites

Ensure that the following libraries are installed for the code to run properly. Install them using the specified version with the following commands:

• npm

  npm install npm@latest -g

• library

  npm install library_name_here -g

Installation

After downloading the project, install the required libraries. In the console, run the command:

streamlit run app.py

(back to top)

Roadmap

• Identify possible datasets
  • Balance the data
  • Create meta.csv (labels)
• Create a preprocessing pipeline
  • Data cleaning and preparation
  • Feature extraction and selection
• Develop the analytical models
  • Implement Root Cause Analysis
  • Develop Sentiment Analysis models
  • Apply Topic Modelling techniques
• Conduct Text Complexity and Readability Analysis
• Perform Geographic Analysis of Complaints
  • Regional Sentiment Analysis
  • Topic Modelling by Geography
• Build Predictive Models for Resolution Outcomes
• Conduct Benchmarking and Comparative Analysis
  • Company Performance Analysis
  • Product Type Analysis
• Deploy the application
  • Cloud deployment
• Future improvements
  • Advanced NLP features
  • Multi-language support

(back to top)

Contact

Your Name - @rishitsaraf - rishitsaraf24@gmail.com

Demo

View a Demo: https://vimeo.com/940595873?share=copy