Research Project · SLIIT · 2025–26 · Research Code: 25_26J_393

AI-Driven Demand Forecasting &
Waste Management System
for Sri Lankan Restaurants

An integrated, end-to-end AI platform that predicts menu demand, optimises ingredient procurement, forecasts food prices, and empowers suppliers purpose-built for Sri Lanka's small-to-medium restaurant ecosystem.

Background & Literature Survey

Small-to medium-sized restaurants in Sri Lanka face chronic operational inefficiencies driven by unpredictable demand, seasonal price volatility, and limited access to data-driven tools. Manual estimation, ad-hoc supplier relationships, and no structured waste tracking result in stock-outs, over-purchasing, and significant food waste — challenges that established solutions from large supply-chain contexts cannot directly address.

01

Time-Series Forecasting

ARIMA and exponential smoothing are well-established for short-term commodity prediction. However, they fail to capture nonlinear demand patterns and Sri Lanka–specific cultural surges.

02

Machine Learning Approaches

Random Forest, Gradient Boosting (LightGBM), and LSTM networks outperform classical methods on volatile, multivariate datasets. Most existing implementations assume large, clean historical data.

03

Supplier Recommendation Systems

Multi-Criteria Decision Making (MCDM) with AHP and ML-based ranking evaluate suppliers by cost, reliability, and freshness — primarily in large supply chains.

04

Privacy-Preserving Forecasting

Federated Learning enables collaborative model training without sharing raw sales data. Differential Privacy and secure multi-party computation supplement this for SME settings.

05

Waste Management & Inventory

EOQ and JIT models optimise static inventory but cannot adapt to perishable, demand-volatile restaurant environments. AI-powered adaptive buffers offer dynamic correction.

06

External Variable Integration

Weather data, cultural calendars, and temporal features significantly improve demand forecast accuracy for hospitality businesses — yet remain underutilised in small-scale settings.

What the Literature Missed

Low-Data Models Absent

Existing forecasting frameworks require large, structured datasets unavailable to typical Sri Lankan restaurants that maintain only manual logbooks.

Siloed Forecasting

Most tools predict demand in isolation without converting forecasts into recipe-mapped ingredient procurement plans, making outputs non-actionable.

No Operational Feedback

Inventory buffers are set statically and never adjusted based on actual kitchen waste — missing a critical closed-loop feedback mechanism.

Privacy Barriers

Restaurant owners resist sharing raw transactional data with suppliers, yet collaborative forecasting requires some form of aggregated demand intelligence.

Our Contribution

An integrated end-to-end system combining lightweight hybrid forecasting, adaptive waste-driven inventory control, price prediction, and privacy-preserving supplier dashboards.

Context-Aware Sri Lanka Focus

Incorporates the Sri Lankan cultural calendar, local weather, and SME operational realities — features no existing solution currently addresses holistically.

"

Research Problem

How can an integrated system be created that forecasts restaurant menu demand, anticipates ingredient price fluctuations, recommends reliable suppliers, and reduces food waste — specifically for Sri Lankan restaurants operating with limited structured data, low digital infrastructure, and privacy-sensitive supply relationships?

What This Project Achieves

  • Develop a hybrid ML + rule-based demand forecasting model tailored for data-scarce Sri Lankan SMEs
  • Integrate cultural calendar and weather data to capture local demand surges accurately
  • Build an adaptive inventory buffer driven by actual kitchen waste via exponential smoothing
  • Predict weekly and monthly food selling prices using ARIMA, Prophet, Regression, and LSTM models
  • Implement a supplier ranking and recommendation engine based on cost, reliability, and freshness
  • Design a privacy-preserving supplier demand forecasting dashboard using federated learning
  • Deliver an intuitive, low-literacy-friendly interface for non-technical restaurant owners
  • Validate accuracy using MAE, RMSE, R², and MAPE against baseline and static-buffer methods

Tech Stack

Python 3.10+LightGBMFacebook Prophet ARIMA / StatsmodelsLSTM / TensorFlowScikit-learn Next.js / ReactDjango REST FrameworkPostgreSQL MongoDBFirebase FirestoreOpenWeatherMap API Pandas / NumPyTailwind CSSXGBoost TypeScriptJoblibFastAPI GitHub

Four Integrated Research Components

Each component is independently developed and seamlessly integrated into a unified decision-support pipeline for Sri Lankan restaurants and their suppliers.

Component 01

Lightweight Hybrid Food Demand Forecaster

Thilakarathna W.P.N.S.  ·  IT22073846  ·  B.Sc. (Hons) IT — SLIIT

The foundational forecasting engine of the system. Designed specifically for small Sri Lankan restaurants with limited historical data, this component combines a lightweight ML core with a weighted fuzzy-logic contextual engine to produce context-aware daily and weekly demand predictions.

The ML core uses LightGBM and Facebook Prophet models trained on historical sales data augmented with engineered temporal features — lag-1, lag-7, and rolling averages. The contextual engine overlays real-time OpenWeatherMap meteorological data and a standardised Sri Lankan cultural calendar (Vesak, Sinhala New Year, Poya Days) to refine predictions.

Unlike binary rule triggers, the system applies Fuzzy Logic Scaling — a continuous adjustment factor (δ) that gradually reduces forecasts as rainfall intensifies rather than switching on/off. Overlapping modifiers (e.g., festival during a storm) are resolved via a Weighted Priority Hierarchy.

FINAL FORECAST FORMULA

Ŷfinal = Ŷbase × (Σ wiδi) / Σ wi

Where Ŷbase = calibrated ML baseline, δi = fuzzy adjustment factor, wi = historical reliability weight

Research Gap Addressed

Existing models assume large, clean data. This component uses proxy data (supplier invoices, logbooks), operates on <1 year of records, and encodes domain expert knowledge through transparent rule-based overrides — making it the first forecasting system tailored to Sri Lankan SME constraints.

SYSTEM ARCHITECTURE — COMPONENT 01

Manual Sales CSV
Supplier Invoices
OpenWeatherMap API
Sri Lankan Cultural Calendar
Data Preprocessing
Cleaning · Feature Engineering · Lag / Rolling
ML Core
LightGBM + Prophet
Contextual Engine
Fuzzy Logic · Weighted Rules
↘   ↙
Final Demand Forecast
Ŷfinal = Ŷbase × Contextual Modifier
ModelMAERMSEMAPE
Proposed Hybrid Model3.424.890.8812.5%
LightGBM (standalone)3.42
Facebook Prophet5.12
LightGBMFacebook ProphetFuzzy Logic Scikit-learnOpenWeatherMapNext.js Django DRFTimeSeriesSplit

Component 02

AI-Enabled Inventory & Reorder Management System

Jayathunga A.G.I.A.  ·  IT22642950  ·  B.Sc. (Hons) IT — SLIIT

This component transforms demand forecasts into physical inventory actions. It computes total ingredient requirements through recipe mapping, dynamically calculates reorder thresholds, and features a pioneering Adaptive Buffer Algorithm that continuously updates safety stock levels using actual kitchen waste data.

The dynamic inventory planner calculates the target stock level as:

TARGET STOCK CALCULATION

Starget = Lreorder + Bsize

Btarget = Wavg × Dbuffer

Bnew = Bold(1 − α) + Btarget · α

α = smoothing factor · Wavg = mean daily waste over lookback window

The adaptive buffer treats kitchen waste as an active data signal rather than a passive metric. By tracking daily unsold food and applying exponential smoothing, the system autonomously corrects safety buffers without human intervention — linking predictive analytics with real kitchen operations.

The Next.js / React frontend presents complex risk states (OK / Low / Stockout) in a simple, visually intuitive dashboard, purpose-designed for restaurant staff with limited digital literacy.

Next.js / ReactTypeScript Django REST FrameworkPostgreSQL JoblibTailwind CSS Exponential SmoothingRecipe Mapping Engine

ADAPTIVE BUFFER FEEDBACK LOOP

Menu Demand Forecast (from Component 01)
Recipe Mapping Engine
Forecast → Ingredient Quantities
Dynamic Inventory Planner
Starget = Lreorder + Bsize
Adaptive Buffer Algorithm
Bnew = Bold(1−α) + Btargetα
Purchase Orders → Supplier
Kitchen Waste Tracking (Daily Input)
↑ Waste feeds back to update buffer automatically
Food Waste Reduced
<5%
Stockout Rate
30d
Simulation Period

Component 03

Food Price Prediction & Supplier Recommendation

Fernando W.G.P.N  ·  IT22261946  ·  B.Sc. (Hons) IT — SLIIT

Most Sri Lankan restaurants rely on manual cost estimation with no predictive capability for ingredient price fluctuations. This component delivers a machine learning–based food price prediction engine and an intelligent supplier ranking and recommendation system — integrated into a unified decision-support interface.

Food Price Prediction Model

Using historical food selling prices, ingredient cost data, and current market prices, the component forecasts weekly and monthly dish selling prices. Models benchmarked include ARIMA, Prophet, Regression, and LSTM — evaluated on RMSE, MAPE, and R².

Feature engineering enriches the dataset with rolling price averages, seasonal indicators, and a computed cost index per dish. Dish-ingredient ratio mapping and supplier-ingredient mapping enable precise cost-to-price linkage.

Supplier Recommendation Engine

A multi-criteria ranking algorithm scores each supplier on cost efficiency, delivery reliability, and freshness ratings. AHP-based MCDM ensures transparent, explainable recommendations. The system alerts restaurant owners when significant price fluctuations are detected and supports PDF/Excel export of supplier rankings.

ARIMAFacebook Prophet LSTM / TensorFlowScikit-learn AHP / MCDMSQLite / MongoDB React (Optional UI)WhatsApp Business API

SYSTEM ARCHITECTURE — COMPONENT 03

Historical Ingredient Data & Market Prices
Supplier Price Rates
Historical Food Sales Dataset
↓ ↓ ↓
Data Preprocessing
Cost-price mapping · Dish-ingredient ratio · Supplier mapping
Price Prediction
ARIMA · Prophet · LSTM · Regression
Supplier Rating ML
Cost · Reliability · Freshness
Optimised Selling Price
Weekly / Monthly
Best Supplier per Ingredient
↘   ↙
Decision Support UI — Suggest Prices · Recommend Suppliers

KEY FUNCTIONAL REQUIREMENTS

  • Weekly and monthly dish price predictions
  • Best supplier ranked per ingredient
  • Price-spike alerts via WhatsApp or notifications
  • Decision-support report generation
  • PDF and Excel export for business use

Component 04

Supplier Demand Forecasting Dashboard

Alawaththa A.K.A.A  ·  IT22249470  ·  B.Sc. (Hons) IT — SLIIT

Small suppliers serving Sri Lankan restaurants lack demand visibility, leading to inefficient delivery scheduling and stock imbalances. This component builds a privacy-preserving supplier demand forecasting dashboard that aggregates ingredient-level demand signals from multiple restaurants without exposing raw sales data.

Forecasting Models

The system collects dish-level sales data from restaurants, converts it to ingredient-level demand via predefined recipes, and applies an ensemble of ARIMA (stationary time-series), Holt-Winters (trend + seasonality), and Random Forest Regression (nonlinear demand) models — weighted by rolling MAPE/RMSE performance.

Privacy-Preserving Aggregation

Using Federated Learning (FL), each restaurant trains its local model and shares only model parameters (gradients) — never raw transactional data. An optional Differential Privacy layer adds controlled noise to further protect individual restaurant identity. This approach enables collaborative forecasting while maintaining competitive confidentiality.

Inventory Optimisation Logic

Supplier-facing reorder points are computed using Economic Order Quantity (EOQ) and safety stock from forecast error. The interactive dashboard (React/Dash) provides time-series demand graphs, inventory heatmaps, stockout/overstock alerts, and drill-down analysis with CSV/PDF export.

ARIMAHolt-Winters Random ForestFederated Learning Differential PrivacyReact / Dash Python / PandasSQL / MongoDB AWS / AzureWhatsApp API

SYSTEM ARCHITECTURE — COMPONENT 04

Restaurant App
Mobile / Web
Local Store
Recipes · Daily Logs
API Gateway & Auth
Forecasting Service
ARIMA · Holt-Winters · RF
Aggregation Service
Privacy-Preserving (FL)
↘   ↙
Data Store
Forecasts · Inventory
Supplier Dashboard (Web)
Forecasts · Heatmaps · Alerts · EOQ
Notification Service
SMS · Email · WhatsApp
Primary Metrics MAE · RMSE · MAPE
Stockout Reduction
Spoilage Reduction
Secondary Metrics Dashboard Satisfaction
Cost Savings / Store
System Reliability

Milestones & Assessments

All four components follow the academic year timeline (September 2025 – August 2026), spanning data collection through final presentation.

Project Proposal

Completed
Date: 08th September 2025 Assessment 01

Individual proposal reports submitted by all four team members. Research gaps identified, methodologies defined, and systems architectures drafted for all four components.

Progress Presentation I

Completed
Date: 09th January 2026 Assessment 02

Data collection completed, preprocessing pipelines built, and initial ML model training underway for all components. Baseline forecasting models benchmarked.

Progress Presentation II

Completed
Date: 9th March 2026 Assessment 03

Full system integration milestone. Hybrid forecaster integrated with inventory management. Supplier dashboard connected with federated learning layer. End-to-end pipeline functional and under evaluation.

Final Assessment & Viva

Upcoming
Date: 4th May 2026 Final Assessment

Final research paper submitted. All four system components validated against holdout datasets. Viva voce conducted.

Project Documents

All project documentation including proposal reports, checklists, and the final research paper. Download links will be updated as documents are finalised.

DC
Available
Project Charter

Initial project charter defining scope, team roles, and research objectives for Project 25_26J_393.

PR
Available
Proposal Report — Thilakarathna W.P.N.S.

IT22073846 · Lightweight Hybrid Food Demand Forecaster component proposal.

PR
Available
Proposal Report — Jayathunga A.G.I.A.

IT22642950 · AI Enabled Inventory Management System

PR
Available
Proposal Report — Fernando W.G.P.N

IT22261946 · Food Price Prediction & Supplier Recommendation component proposal.

PR
Available
Proposal Report — Alawaththa A.K.A.A

IT22249470 · Supplier Demand Forecasting Dashboard component proposal.

CL
In Progress
Checklist Documents

Assessment checklists and milestone compliance documents for all project phases.

RD
In Progress
Research Paper — Draft V1

Combined research paper covering Components 01 & 02 (IEEE format). Currently under revision.

FR
Pending
Final Research Document (All 4 Components)

Complete final research document integrating all four components — due August 2026.

Slides of Past Presentations

Presentation decks from all assessment milestones. Future presentations will be added here upon completion.

01

Proposal Presentation

Initial research proposal presentation covering all four components, research gaps, and methodology overview.

Completed 08th September 2025
02

Progress Presentation I

Interim results from data collection, preprocessing pipelines, and initial model training across all components.

Completed 7th January 2026
03

Progress Presentation II

Full integration showcase — hybrid forecaster pipeline, adaptive buffer algorithm live demo, and supplier dashboard.

Completed 9th March 2026
04

Final Presentation

Comprehensive final results, system evaluation metrics, waste reduction outcomes, and business impact analysis.

Upcoming 4th May 2026

About Us

Final year undergraduate students at the Sri Lanka Institute of Information Technology (SLIIT), pursuing B.Sc. (Hons) in Information Technology. Project ID: 25_26J_393.

J

Jayathunga A.G.I.A.

IT22642950
AI Supported Inventory and Reorder Management

Research Paper Primary Author · Adaptive Buffer Algorithm

T

Thilakarathna W.P.N.S.

IT22073846
Demand Forecasting and Rule Engine

Hybrid ML + Rule-Based Forecasting · LightGBM, Prophet

F

Fernando W.G.P.N.

IT22261946
Price Prediction and Supplier Recommendation

ARIMA · LSTM · AHP Supplier Ranking · Decision Support

A

Alawaththa A.K.A.A.

IT22249470
Supplier Dashboard and Demand Forecasting

Supplier Demand Forecasting · Supplier Dashboard · EOQ

S

Mr. Ravi Supunya

Project Supervisor · Department of Information Technology, SLIIT

supunya.s@sliit.lk

S

Mrs. Chathurya Kumarapperuma

Co-Supervisor · Department of Information Technology, SLIIT

chathurya.k@sliit.lk

Contact Us

Reach Out

Have questions about our research? We'd love to hear from you.

Thilakarathna W.P.N.S.

it22073846@my.sliit.lk · 0719355359

Jayathunga A.G.I.A.

it22642950@my.sliit.lk · 0759089188

Alawatta A.K.A.A.

it22249470@my.sliit.lk · 0764233231

Fernando W.P.G.N.

it22261946@my.sliit.lk · 0776474109