AI-Powered Skincare Recommendation System

https://corium.streamlit.app/

An interactive Machine Learning web application that provides personalized skincare product recommendations and daily skincare routines based on a user's skin profile, concerns, and severity levels.

Built using Streamlit and Scikit-Learn, the application demonstrates the complete machine learning workflow including dataset generation, data preprocessing, model training, evaluation, and deployment.

Features

• Personalized skincare recommendations
• Skin health score calculation
• Morning and night routine generation
• Recommendation history tracking
• Downloadable PDF reports
• Interactive ML Insights dashboard
• Random Forest based recommendation engine
• Synthetic dataset generation for model training

Machine Learning Workflow

1. Dataset Generation

A synthetic skincare dataset containing 1,200+ records is generated using different skin types, concerns, severity levels, and recommended skincare products.

2. Data Preprocessing

• Data cleaning
• Feature encoding
• Label transformation
• Training and testing split

3. Model Training

A Random Forest model is trained to predict suitable skincare products based on:

• Age
• Skin Type
• Skin Concern
• Severity Score

4. Model Evaluation

The model is evaluated using:

• Accuracy
• Precision
• Recall
• F1 Score
• Confusion Matrix
• Feature Importance Analysis

5. Recommendation Generation

The trained model predicts suitable:

• Cleanser
• Moisturizer
• Serum
• Sunscreen

The recommendations are then combined with rule-based routine generation to create personalized skincare plans.

Tech Stack

Frontend

• Streamlit

Machine Learning

• Scikit-Learn
• Pandas
• NumPy
• Joblib

Visualization

• Matplotlib

Reporting

• ReportLab

Project Structure

skincare-recommendation/
│
├── data/
├── ml/
├── models/
├── pages/
├── products/
├── tests/
├── utils/
│
├── app.py
├── requirements.txt
└── README.md

Installation

Clone the repository:

git clone https://github.com/your-username/ai-skincare-recommendation-system.git
cd ai-skincare-recommendation-system

Install dependencies:

pip install -r requirements.txt

Generate the dataset:

python ml/generate_dataset.py

Train the model:

python ml/train_model.py

Run the application:

streamlit run app.py

Future Improvements

• Real skincare product database integration
• User authentication and profiles
• Skin image analysis using computer vision
• Deep learning recommendation models
• Cloud deployment and API integration

Author

Gaurang Kishore

Developed to help people better understand their skin concerns and receive personalized skincare recommendations, while applying machine learning techniques to create an accessible and user-friendly skincare guidance system.