DEPI-Graduation-Project

Data Engineering ALX1_AIS4_S1e

Skincare Product Sentiment Analysis System 🧴✨

Team Members:

• Nada Hamdy Fatehy Abedelsalam
• Toqa Mohsen
• Shahd Ammar
• Omar Salah
• Yousef Magdy

Project Overview

This project is a collaborative effort aimed at analyzing customer feedback for skincare products and predicting the sentiment (positive or negative) for each review using an unsupervised model. The system includes a web application where users can input reviews and get real-time sentiment predictions based on a pre-trained unsupervised model for sentiment analysis.

Key Features

• 📊 Sentiment analysis for skincare product reviews using an unsupervised model.
• 🌐 Web application for real-time sentiment predictions.
• ⚡ ETL pipeline built using Azure Data Factory, Azure Databricks, and Azure Synapse Analytics.
• 🚀 Optimized for large-scale data processing with Azure services.

Tech Stack

• Azure Data Factory for ETL orchestration
• Azure Databricks for data processing
• Azure Synapse Analytics for data storage and analysis
• Unsupervised model for sentiment analysis
• Flask for the web application

Table of Contents

1. Setup
2. ETL Pipeline
3. Model Details
4. Website
5. Power BI Dashboard
6. Contributing
7. License

Setup

Prerequisites

• Python 3.12 or higher
• Azure Subscription (can be student subscription)
• Access to Azure Data Factory, Databricks, and Synapse Analytics
• Flask for running the web app

Repository Access

• Ensure all team members have access to the shared repository.
• Collaborate using branches for feature development.

ETL Pipeline

The ETL pipeline is designed to handle large volumes of customer feedback data:

• Azure Data Factory: Ingests raw review data from various sources.

• Azure Databricks: Processes and cleans the data using predefined transformations.

• Azure Synapse Analytics: Stores processed data for analysis and visualization in Power BI.

Model Details

This project leverages a pre-trained unsupervised sentiment analysis model to classify product reviews as positive or negative. The model is used in the web app for internal real-time predictions but is not designed for API usage or external requests.

Website

The web application allows users to submit product reviews and instantly receive sentiment predictions based on the unsupervised model.

Key Features:

• Simple and intuitive interface for entering product reviews.
• Displays the predicted sentiment of the review.
• Deployed using Flask.

Power BI Dashboard

A Power BI dashboard is used for advanced data visualization and analysis, allowing users to explore trends in review data:

• Sentiment distribution for skincare products.

• Top-rated products based on customer feedback.

• Time-based analysis of reviews.

• Dashboard Preview

Documentation

• All the details are recorded here Google Docs

Technologies Used

• Python: Backend logic and machine learning model.
• Flask: Web framework for building the backend API.
• HTML/CSS/JavaScript: Frontend interface.
• Azure: Azure Blob Storage and Azure App Service for deployment.

Contributing

This is a collaborative project. To contribute:

1. Work on your feature or bug fix in a separate branch.
2. Ensure your changes are tested and reviewed by another team member.
3. Submit a pull request when your work is ready for review.