Inventory Demand Forecasting

A full-stack inventory planning application built with Django REST Framework and React. It tracks daily consumption, monitors stock health, and runs machine learning forecasts to support reorder decisions.

Highlights

• JWT-based authentication with role-based access (admin, analyst)
• Master data modules for items, stocks, suppliers, and users
• Consumption logging with automatic stock adjustment
• Forecasting pipeline with model comparison (Linear Regression, Random Forest, Prophet)
• Dashboard with KPI cards, priority queue, and visual charts (bar + donut)
• Forecast screen with confidence bands and CSV export
• Clean frontend service layer using axios via a shared API client

Tech Stack

Backend

• Django 5.2.12
• Django REST Framework 3.17.1
• SimpleJWT 5.5.1
• SQLite (dev), PostgreSQL-compatible config for production
• scikit-learn, pandas, numpy, Prophet

Frontend

• React 19.2.4
• React Router 7.14.0
• Recharts 3.8.1
• Axios 1.14.0
• Vite 5.4.x

Repository Structure

Inventory_demand_forecasting/
|- backend/
|  |- accounts/                # custom user model + auth APIs
|  |- consumption/             # consumption CRUD + stock sync logic
|  |- dashboard/               # aggregated dashboard APIs
|  |- forecasting/             # forecast APIs + ML pipeline
|  |  \- ml/
|  |- inventory/               # items and stocks
|  |- suppliers/               # supplier CRUD
|  |- utils/management/commands/
|  |  |- load_synthetic_data.py
|  |  \- run_forecasts.py
|  |- resources/data/          # synthetic dataset generator + JSON files
|  |- core/                    # settings + root urls
|  \- manage.py
\- frontend/
   |- src/
   |  |- components/
   |  |  \- dashboard/DashboardCharts.jsx
   |  |- context/
   |  |- hooks/
   |  |- layouts/
   |  |- pages/
   |  |- routes/
   |  |- services/             # separated API modules, shared apiClient
   |  \- styles/
   |- .env.example
   \- package.json

Functional Modules

• Landing page: Overview, Features, Workflow sections
• Dashboard: KPIs, top consumption chart, stock status donut, supplier value chart, action queue
• Forecasts: model comparison + forecast confidence visualization + CSV export
• Items and Stocks: full inventory lifecycle
• Suppliers: directory and edits
• Consumption: create/read/update/delete with stock impact
• User Management: admin creates analysts

Screenshots

The following screenshots show the main operational modules in the app UI.

Dashboard

Shows KPI cards, action-ready stock alerts, and visual charts for demand, stock status, and supplier exposure.

Consumption

Capture daily consumption entries and review recent usage history with filters.

Forecasts

Model-driven demand forecast view with horizon selection, confidence bands, and reorder guidance.

Items

Item master data management including search/filter and admin-only create/update controls.

Stocks

Stock threshold management for reorder level, safety stock, and max stock settings.

Suppliers

Supplier directory with contact details, lead times, and minimum order quantity.

Users

Admin-only analyst account creation and analyst user listing.

Access Control

Capability	Admin	Analyst
View dashboard/forecasts/items/stocks/suppliers/consumption	Yes	Yes
Create analysts	Yes	No
Create/update/delete items	Yes	No
Update stocks	Yes	No
Create consumption entries	Yes	Yes
Update/delete consumption entries	Yes	No
Create/update/delete suppliers	Yes	No

API Overview

Base URL (default): http://127.0.0.1:8000/api

Accounts (/api/accounts/)

Method	Endpoint	Description	Access
POST	/login/	Login and get access + refresh token	Public
POST	/logout/	Blacklist refresh token	Authenticated
GET	/analysts/	List analysts	Admin
POST	/analysts/	Create analyst	Admin
GET	/test/	JWT test endpoint	Authenticated

Suppliers (/api/suppliers/)

Method	Endpoint	Access
GET	/	Admin + Analyst
GET	/:id/	Admin + Analyst
POST	/	Admin
PUT/PATCH	/:id/	Admin
DELETE	/:id/	Admin

Inventory (/api/items/, /api/stocks/)

Method	Endpoint	Access
GET	/items/, /items/:id/	Admin + Analyst
POST/PUT/PATCH/DELETE	/items/, /items/:id/	Admin
GET	/stocks/, /stocks/:id/	Admin + Analyst
POST/PUT/PATCH/DELETE	/stocks/, /stocks/:id/	Admin

Consumption (/api/consumption/)

Method	Endpoint	Access
GET	/, /:id/	Admin + Analyst
POST	/	Admin + Analyst
PUT/PATCH/DELETE	/:id/	Admin

Dashboard (/api/dashboard/)

Method	Endpoint	Description
GET	/stock-overview/	stock status snapshot
GET	/consumption-summary/?days=7	per-item consumption summary
GET	/low-stock-alerts/	reorder-risk items
GET	/supplier-summary/	supplier-level value summary

Forecasting (/api/forecasting/)

Method	Endpoint	Description
GET	/results/?days=30	forecast rows for horizon
GET	/results/?item_id=101&days=30	forecast rows for one item
GET	/performance/	model performance list
GET	/performance/compare/?item_id=101	best model + per-model metrics

Forecasting Pipeline

Run from backend/:

python manage.py run_forecasts

Useful options:

python manage.py run_forecasts --item-id 101
python manage.py run_forecasts --days 30

What it does:

1. Loads historical consumption by item
2. Builds engineered features (date, lag, rolling, holidays)
3. Trains 3 models
4. Evaluates MAPE/RMSE/MAE
5. Marks best model per item
6. Saves forecasts to ForecastResult

Local Setup

1) Clone

git clone https://github.com/HarshAI-ML/Inventory_demand_forecasting.git
cd Inventory_demand_forecasting

2) Backend setup

cd backend
python -m venv venv

# Windows
venv\Scripts\activate

# macOS/Linux
source venv/bin/activate

pip install -r requirements.txt
python manage.py migrate
python manage.py createsuperuser
python manage.py runserver

Backend default: http://127.0.0.1:8000

3) Frontend setup

cd frontend
npm install
npm run dev

Frontend default: http://127.0.0.1:5173

Environment Variables

Backend (backend/.env)

ENV=development
DJANGO_SECRET_KEY=change-me
DJANGO_DEBUG=True
DJANGO_ALLOWED_HOSTS=localhost,127.0.0.1

# Used when ENV=production
DB_NAME=
DB_USER=
DB_PASSWORD=
DB_HOST=
DB_PORT=5432

Frontend (frontend/.env)

VITE_API_BASE_URL=http://127.0.0.1:8000/api

Synthetic Data

From backend/:

python resources/data/generate_data.py
python manage.py load_synthetic_data

# optional reset + reload
python manage.py load_synthetic_data --flush

Frontend API Layer

The frontend keeps concerns separated:

• src/services/apiClient.js: shared axios request wrapper, token/header handling, error normalization
• src/services/*.js: domain services (authService, dashboardService, inventoryService, etc.)

Database Schema

erDiagram
  Supplier ||--o{ Item : "supplies"
  Item ||--|| InventoryStock : "tracked in"
  Item ||--o{ Consumption : "consumed as"
  Item ||--o{ ForecastResult : "forecasted for"
  Item ||--o{ ModelPerformance : "evaluated for"

  User {
    int id PK
    string username
    string email
    string role
    string phone
    string department
  }
  Supplier {
    int supplier_id PK
    string supplier_name
    string location
    int lead_time
    int minimum_order_quantity
  }
  Item {
    int item_id PK
    string item_name
    string category
    int supplier_id FK
    decimal cost_per_unit
    int shelf_life_days
  }
  InventoryStock {
    int stock_id PK
    int item_id FK
    int quantity_available
    int reorder_level
    int safety_stock
    int max_stock_level
  }
  Consumption {
    int consumption_id PK
    int item_id FK
    int quantity_used
    date date
    string department
    bool holiday_flag
  }
  ForecastResult {
    int id PK
    int item_id FK
    date forecast_date
    float predicted_qty
    float lower_bound
    float upper_bound
    string model_used
  }
  ModelPerformance {
    int id PK
    int item_id FK
    string model_name
    float mape
    float rmse
    bool is_best_model
  }

Loading

Useful Commands

# backend tests
cd backend
python manage.py test

# frontend lint
cd frontend
npm run lint

# frontend production build
npm run build