FeedbackIQ

Transform Customer Reviews into Actionable Insights with AI

FeedbackIQ is an AI-powered review intelligence platform that analyzes large-scale customer feedback and converts raw reviews into meaningful business insights using Natural Language Processing (NLP).

---

Demo

Add your project GIF here once available.

assets/
└── feedbackiq-demo.gif

![FeedbackIQ Demo](assets/feedbackiq-demo.gif)

---

Problem Statement

Modern businesses receive thousands of customer reviews across marketplaces, applications, and digital platforms.

Manually analyzing this feedback is:

• Time-consuming
• Expensive
• Error-prone
• Difficult to scale

FeedbackIQ automates this process by leveraging machine learning and topic modeling to identify customer sentiment, uncover recurring themes, and generate actionable insights within minutes.

---

Features

CSV-Based Review Analysis

Upload customer review datasets directly through a simple interface.

Sentiment Analysis

Classify reviews and understand overall customer satisfaction.

• Positive Reviews
• Negative Reviews
• Neutral Reviews

Topic Extraction

Discover recurring customer concerns and feedback patterns using BERTopic.

Examples:

• Battery Life Issues
• Customer Support Complaints
• Delivery Delays
• Product Quality Concerns

Interactive Analytics Dashboard

Visualize:

• Sentiment Distribution
• Review Statistics
• Topic Insights
• Historical Reports

Report Management

• View Previous Reports
• Access Historical Analysis
• Delete Reports
• Compare Review Data Over Time

Statistics Endpoint

Track platform-level metrics:

• Total Reports Generated
• Total Reviews Analyzed
• Average Positive Sentiment

---

How It Works

User Uploads CSV
        │
        ▼
Frontend (React)
        │
        ▼
Backend API (Express.js)
        │
        ▼
ML Service (FastAPI)
        │
 ┌──────┴──────┐
 ▼             ▼
Sentiment   Topic
Analysis   Extraction
(Logistic   (BERTopic)
Regression)
        │
        ▼
MongoDB
        │
        ▼
Analytics Dashboard

---

Architecture

┌─────────────────────┐
│     React App       │
└──────────┬──────────┘
           │
           ▼
┌─────────────────────┐
│   Express Backend   │
└──────────┬──────────┘
           │
           ▼
┌─────────────────────┐
│  FastAPI ML Service │
└──────────┬──────────┘
           │
           ▼
┌─────────────────────┐
│      MongoDB        │
└─────────────────────┘

---

Tech Stack

Frontend

• React
• Vite
• Axios
• React Router
• Modern Responsive UI

Backend

• Node.js
• Express.js
• MongoDB
• Mongoose
• Multer

Machine Learning

• FastAPI
• Scikit-Learn
• TF-IDF Vectorization
• Logistic Regression
• BERTopic
• Pandas
• NumPy

---

Machine Learning Pipeline

Sentiment Analysis

The sentiment engine is built using:

• Text Preprocessing
• TF-IDF Vectorization
• Logistic Regression Classifier

Model Performance

Model	Accuracy
Logistic Regression	88.54%

Training Dataset

The model was trained on a custom combined dataset sourced from:

• Amazon Reviews
• IBM Customer Feedback
• Google Play Store Reviews

This combination improves the model's ability to generalize across multiple review domains.

---

Topic Modeling

FeedbackIQ uses BERTopic to identify recurring themes from customer feedback.

Example:

Input Reviews

Battery drains quickly.
Battery backup is poor.
The phone only lasts a few hours.

Generated Topic

Battery Life Issues

---

CSV Format

Your CSV file must contain a column named:

review
Amazing product
Battery life is terrible
Customer support was very helpful

Important

The column name must be exactly:

review

Not:

Review
REVIEWS
customer_review
feedback

---

API Endpoints

Upload Reviews

POST /upload

Uploads and validates CSV data.

---

Create Report

POST /reports

Creates a new analysis report.

---

Get All Reports

GET /reports

Returns all generated reports.

---

Get Report By ID

GET /reports/:id

Returns detailed report information.

---

Delete Report

DELETE /reports/:id

Removes a report.

---

Statistics

GET /reports/stats

Example Response:

{
  "totalReports": 12,
  "totalReviewsAnalyzed": 15432,
  "averagePositivePercentage": 78.4
}

---

Project Structure

FEEDBACKIQ/
│
├── frontend/
│
├── backend/
│
├── ml/
│
├── assets/
│   └── feedbackiq-demo.gif
│
└── README.md

---

Installation

Clone Repository

git clone https://github.com/alwaysphoenixR/FEEDBACKIQ.git
cd FEEDBACKIQ

---

Frontend Setup

cd frontend
npm install
npm run dev

---

Backend Setup

cd backend
npm install
npm run dev

---

ML Service Setup

cd ml

pip install -r requirements.txt

uvicorn app.main:app --reload

---

Environment Variables

Backend

PORT=
MONGO_URI=
ML_SERVICE_URL=

ML Service

MODEL_PATH=

---

User Experience

FeedbackIQ is designed to make large-scale review analysis effortless.

While reviews are being processed, users are shown a friendly waiting screen:

“Analyzing reviews... This may take a while. Brew a coffee ☕ while FeedbackIQ works.”

---

Future Improvements

• Aspect-Based Sentiment Analysis
• Multi-language Support
• Comparative Report Analysis
• PDF Report Export
• AI-Powered Recommendations
• Trend Analysis Over Time
• Real-Time Review Monitoring
• LLM-Based Insight Generation

---

Why FeedbackIQ?

FeedbackIQ demonstrates:

Frontend Engineering

• React Development
• API Integration
• Responsive Dashboard Design

Backend Engineering

• REST API Development
• File Upload Handling
• Database Design
• Service-Oriented Architecture

Machine Learning

• NLP Pipelines
• Sentiment Analysis
• Topic Modeling
• FastAPI Deployment

System Design

• Frontend ↔ Backend ↔ ML Communication
• Multi-Service Architecture
• Production-Oriented Design

---

Author

Rajveer Singh

Computer Science & Engineering
NIT Raipur

---

Star the Repository

If you found this project useful, consider giving it a star.

It helps support the project and motivates future improvements.