🧪 Smart AI Shelf-Life Prediction Platform

By Ajinkya Ghuge

An AI-powered intelligence platform built to predict the expiry dates and shelf stability of food and pharmaceutical products using machine learning.

"No more manual tracking or guesswork. Just smart, accurate stability predictions based on data."

---

📋 Table of Contents

#	Section
1	Project Overview
2	Key Features
3	System Architecture
4	Machine Learning Pipeline
5	Project Workflow
6	Tech Stack
7	Project Structure
8	Installation & Local Setup
9	API Reference

---

🎯 Project Overview

Manual tracking and guesswork regarding product expiration lead to massive inventory losses, safety risks, and operational waste. The Expiry Date Predictor acts as an automated solution that evaluates intricate raw feature variables—such as dynamic ingredients, volatile packaging materials, and active storage environment conditions—to forecast an exact, data-backed shelf life duration matrix.

Core Objectives

• 🧾 Minimize Resource Waste: Assists production facilities in mitigating supply chain decay through preventative timelines.
• 🛡️ Optimize Safety Assurance: Provides chemical, pharmaceutical, and consumable goods providers with predictive safety safety nets.
• 📦 Streamline Inventory Flow: Equips warehouse managers with accurate predictive indices for first-expired, first-out (FEFO) strategies.

---

✨ Key Features

• 🔍 Multi-Variable Feature Analysis: Aggregates multi-dimensional product configurations (e.g., compounding components, air exposure profiles, baseline temperature bounds).
• 🧠 Advanced Regression Inference: Uses trained scikit-learn estimators to convert dynamic environmental thresholds into real-time shelf life counts.
• 📊 Insightful Exploratory Dashboards: Packaged alongside exploratory rendering matrices utilizing Matplotlib and Seaborn for dataset profiling.
• 🔌 API Integration Layer: Exposes modular HTTP service routes designed to accept payload vectors and stream instantaneous predictions out to client ERP nodes.

---

🏗️ System Architecture

Component Diagram

flowchart TB

    subgraph ClientLayer["User Web Interface Layer"]
        A["HTML Input Dashboard<br/>Product Parameter Form UI"]
        B["Static Assets<br/>Validation Engine"]
    end

    subgraph ServerLayer["Flask Backend"]
        C["app.py<br/>HTTP Route Handlers"]
    end

    subgraph CoreEngine["Predictive AI Layer"]
        D["model.py<br/>ML Inference Engine"]
    end

    subgraph AnalyticsLayer["Analytics & Validation"]
        E["Jupyter Notebooks<br/>EDA & Model Evaluation"]
    end

    A --> B
    B -->|JSON Request| C
    C -->|Prediction Request| D
    D -->|Prediction Response| C
    E -->|Model Validation| D
    C -->|JSON Response| B

Loading

---

🔄 Request Flow

sequenceDiagram

    actor User

    participant Browser
    participant FlaskApp
    participant Model
    participant Analytics

    User->>Browser: Enter Product Information

    Browser->>FlaskApp: POST /predict

    FlaskApp->>Model: Preprocess Features

    activate Model

    Model->>Model: Feature Engineering
    Model->>Model: Run ML Prediction

    Model-->>FlaskApp: Predicted Shelf Life

    deactivate Model

    FlaskApp->>Analytics: Log Prediction Metrics

    Analytics-->>FlaskApp: Tracking Information

    FlaskApp-->>Browser: JSON Response

    Browser-->>User: Display Expiry Prediction

Loading

---

🤖 Machine Learning Pipeline

flowchart LR

    A["Data Collection"] -->
    B["Data Cleaning & Preprocessing"]

    B -->
    C["Feature Engineering"]

    C -->
    D["Model Training"]

    D -->
    E["Model Evaluation"]

    E -->
    F["Expiry Date Prediction"]

Loading

---

🚀 Project Workflow

flowchart TD

    Start([Start])

    --> Load[Load Model]

    Load --> Validate{Model Available?}

    Validate -->|No| Error[Raise Error]

    Validate -->|Yes| Run[Start Flask Server]

    Run --> Wait[Wait For User Request]

    Wait --> Input[Receive Product Details]

    Input --> Process[Preprocess Data]

    Process --> Predict[Generate Prediction]

    Predict --> Result[Create Response]

    Result --> Display[Show Prediction]

    Display --> End([Finish])

Loading

---

🛠️ Tech Stack

Category	Technology	Purpose
Language	Python 3.8+	Core Development
Backend	Flask	Web Application
Data Processing	Pandas, NumPy	Data Manipulation
Machine Learning	Scikit-Learn	Prediction Models
Visualization	Matplotlib, Seaborn	Analytics
Experimentation	Jupyter Notebook	Model Training

---

📂 Project Structure

expiry-predictor/
│
├── data/
│   └── datasets/
│
├── static/
│   ├── css/
│   ├── js/
│   └── images/
│
├── templates/
│
├── notebooks/
│
├── app.py
├── model.py
├── constraint.py
├── requirements.txt
└── README.md

---

⚡ Installation & Local Setup

Prerequisites

• Python 3.8+
• pip

Clone Repository

git clone https://github.com/Ajinkya-Ghuge/Datasphere.git

cd Datasphere

Create Virtual Environment

python -m venv venv

Windows:

venv\Scripts\activate

Linux/macOS:

source venv/bin/activate

Install Dependencies

pip install -r requirements.txt

Run Application

python app.py

Open:

http://localhost:5000

---

🔌 API Reference

Method	Endpoint	Description
GET	/	Load Dashboard
POST	/predict	Predict Expiry Date

Example Request

{
  "temperature": 25,
  "humidity": 60,
  "packaging": "Plastic",
  "ingredients": "Milk Powder"
}

Example Response

{
  "predicted_shelf_life_days": 365,
  "confidence_score": 0.94
}

---

👤 Author

Ajinkya Ghuge

GitHub: https://github.com/Ajinkya-Ghuge

⭐ Star this repository if you found it useful!