🚀 Adaptive Procurement Intelligence System (APIS)

An AI-powered procurement intelligence platform that helps organizations monitor procurement performance, evaluate supplier reliability, detect anomalies, and predict delivery delays using Machine Learning + Analytics.

APIS is built as a multi-page Streamlit dashboard with automated risk scoring, supplier segmentation, anomaly detection, reporting, and model retraining.

📌 Problem Statement

In real-world procurement operations, organizations face challenges such as:

Late deliveries impacting production timelines

Quality issues (high defect rates) increasing costs

Price fluctuations affecting budgeting

Difficulty identifying risky suppliers early

Manual monitoring & lack of intelligent decision support

Traditional procurement systems are reactive. ✅ APIS makes procurement proactive using AI-driven risk intelligence.

🎯 Project Objectives

✔ Monitor procurement order performance in real-time ✔ Score supplier risk from delivery + quality + pricing patterns ✔ Predict order delay probability using supplier history + order parameters ✔ Detect anomalies/outliers automatically (risk, defects, price spikes) ✔ Segment suppliers into reliability groups using clustering ✔ Provide downloadable reports & analytics dashboards ✔ Support adaptive learning through model retraining & training logs

✨ Final Phase Upgrades (All-in-One)

This final version of APIS includes the following major upgrades:

✅ 1) Multi-Page Streamlit Dashboard (Full Navigation)

APIS is now a complete dashboard suite, not just a single page.

Modules included:

🏠 Main Dashboard Home

📊 Overview Analytics

📦 Orders Explorer

🏢 Supplier Intelligence

🧩 Supplier Segmentation (Clustering)

🚨 Alerts & Anomalies

🤖 Delay Predictor (ML)

📄 Reports & Downloads

🔁 Retrain & Logs

Dashboard entry page: _🏠_Dashboard.py

_🏠_Dashboard

✅ 2) Modern Dark Mode UI Theme (Professional Look)

A custom dark theme UI is applied across all pages for:

Better readability

High contrast text

Consistent UI styling

Modern enterprise dashboard feel

Theme is applied using: apply_dark_theme() from theme.py

theme

✅ 3) Supplier Risk Scoring + Ranking System

Suppliers are evaluated using risk scores and categorized into:

🟢 Low Risk

🟠 Medium Risk

🔴 High Risk

The Suppliers module shows:

risk distribution

risk ranking table

progress-style risk bars

Supplier page: 3_🏢_Suppliers.py

3_🏢_Suppliers

✅ 4) Orders Explorer with Advanced Filtering

A full orders explorer is added to filter and analyze orders by:

supplier

order status (OnTime / Delayed)

priority (Low / Medium / High)

sorting by delay, defect rate, quantity, price

Orders Explorer page: 2_📦_Orders_Explorer.py

2_📦_Orders_Explorer

✅ 5) Supplier Segmentation using K-Means Clustering

Suppliers are segmented into strategic categories:

🟢 Reliable

🟠 Moderate

🔴 Risky

This helps procurement teams with:

supplier strategy planning

vendor relationship management

risk mitigation

Clustering page: 4_🧩_Supplier Segmentation.py

4_🧩_Supplier Segmentation

✅ 6) Alerts & Anomaly Detection Dashboard

APIS detects anomalies like:

🚨 High Risk Orders (risk score ≥ 70)

⚠️ Quality issues (defect rate ≥ 6%)

📈 Price spikes (±10% change)

⏳ Delivery delays

Alerts module: 5🚨_Alerts&_Anomalies.py

5🚨_Alerts&_Anomalies

Isolation Forest anomaly engine: anomaly_detection.py

anomaly_detection

✅ 7) ML Delay Prediction (Supplier History Based)

Delay prediction is upgraded to be more realistic using supplier historical performance:

supplier_avg_delay_days

supplier_avg_defect_rate

supplier_on_time_rate

User selects a supplier and enters order details → model predicts:

✅ On-Time 🚨 Delayed

Delay predictor UI: 6_🤖_Delay_Predictor.py

6_🤖_Delay_Predictor

✅ 8) Model Retraining + Best Model Selection (LR vs RF)

APIS supports adaptive learning through retraining:

Trains both Logistic Regression & Random Forest

Selects best model using F1-score

Saves best model to models/model.pkl

Stores training logs to logs/training_log.csv

Generates reports/model_comparison.csv

Retraining engine: retrain_model.py

retrain_model

Retrain dashboard page: 8🔁_Retrain&_Logs.py

8🔁_Retrain&_Logs

✅ 9) Reports & Bulk Export Downloads (ZIP Support)

A dedicated Reports Center allows users to download:

Supplier Risk Report

Anomaly Detection Report

Supplier Clustering Report

Procurement Summary

It also supports bulk export ZIP containing all CSVs.

Reports page: 7📄_Reports&_Downloads.py

7📄_Reports&_Downloads

🏗️ System Workflow / Architecture

1. Order Data Collection (CSV dataset) ⬇
2. Risk Score Calculation + Supplier Ranking ⬇
3. Supplier Clustering (Segmentation) ⬇
4. Anomaly Detection (Isolation Forest) ⬇
5. ML Delay Prediction (Supplier + Order Features) ⬇
6. Dashboard Visualization + Reports Export ⬇
7. Retraining & Logs (Adaptive Learning)

🧰 Tech Stack 🖥️ Frontend / UI

Streamlit (Multi-page dashboard)

⚙️ Backend / Analytics / ML

Python

Pandas / NumPy

Scikit-learn

Joblib

📦 Models Used

RandomForestClassifier

LogisticRegression

IsolationForest

K-Means (for clustering report)

📂 Project Structure

adaptive-procurement-intelligence-system/ │ ├── app/ │ ├── theme.py │ ├── utils.py │ ├── 🏠_Dashboard.py │ └── pages/ │ ├── 1📊Overview.py │ ├── 2📦Orders_Explorer.py │ ├── 3🏢Suppliers.py │ ├── 4🧩Supplier Segmentation.py │ ├── 5🚨Alerts&Anomalies.py │ ├── 6🤖Delay_Predictor.py │ ├── 7📄Reports&Downloads.py │ └── 8🔁Retrain&_Logs.py │ ├── dataset/ │ ├── orders.csv │ ├── suppliers.csv │ ├── supplier_risk_report.csv │ ├── supplier_clusters.csv │ └── anomaly_report.csv │ ├── models/ │ └── model.pkl │ ├── src/ │ ├── retrain_model.py │ ├── anomaly_detection.py │ └── risk_score.py │ ├── reports/ │ └── model_comparison.csv │ ├── logs/ │ └── training_log.csv │ ├── requirements.txt └── README.md

📊 Dataset Details

This project uses procurement order records containing supplier and order performance information.

Common dataset columns include:

order_id

supplier_id

quantity

unit_price

defect_rate

delay_days

order_status (OnTime / Delayed)

order_priority

region

price_change_percent

shipping_mode

payment_terms

item_category

⚙️ Installation & Setup ✅ 1) Clone the Repository git clone https://github.com/RaKa8904/adaptive-procurement-intelligence-system.git cd adaptive-procurement-intelligence-system

✅ 2) Install Dependencies pip install -r requirements.txt

✅ 3) Run the Dashboard streamlit run _🏠_Dashboard.py

🚀 How to Use the Dashboard

Once the dashboard is running:

🏠 Dashboard Home

Shows quick system status, alerts, trends, and top risky suppliers.

KPIs + order distribution + priority breakdown.

📦 Orders Explorer

Filter orders and view performance metrics.

🏢 Suppliers

Supplier master data + risk ranking + risk distribution.

🧩 Supplier Segmentation

Cluster suppliers into reliable/moderate/risky groups.

🚨 Alerts & Anomalies

Detect risky orders, quality issues, price spikes, delays.

🤖 Delay Predictor

Predict whether a new order will be delayed using ML.

📄 Reports & Downloads

Download CSV reports + bulk ZIP export.

🔁 Retrain & Logs

Retrain ML models and monitor training logs.

📌 Future Scope (Optional Enhancements)

Add PostgreSQL/SQL database integration

Role-based authentication (Admin / Analyst)

Real-time supplier alerts via Email/SMS

PowerBI/Tableau connector exports

Explainable AI (SHAP) for prediction reasoning

Deployment on Streamlit Cloud / AWS / Azure

👨‍💻 Author

Built by Raka 💙