Supervisor

Dr. Muhammad Iqbal

Programme

BSc (Hons) in Computing in IT

Subject

Computer Science

Abstract

Flooding costs the Irish economy more than 300 million euros annually; however, the hydrometric and meteorological data needed to predict dangerous river rises is already collected and freely available. This project applies machine learning to a decade of hourly observations from OPW Station 09001 on the River Liffey at Leixlip and Met Eireann Casement Aerodrome to build a flood risk prediction system with a 72-hour forecast horizon.

A two-phase modelling pipeline was built following the CRISP-DM framework. Phase 1 compared LSTM and GRU recurrent neural networks on the task of forecasting the river level in metres 72 hours ahead. LSTM was selected as winner with an RMSE of 0.1148 metres and R2 of 0.640, compared to a persistence baseline R2 of 0.580. Phase 2 used the LSTM forecast as a stacked feature alongside 34 engineered temporal variables to compare Random Forest and XGBoost classifiers on three-class flood risk categorization. Random Forest was selected as winner with an F1-macro of 0.8027 and lower cross-validation variance.

A probabilistic output layer converts Random Forest probabilities into four named warning levels. Evaluated against 2,994 actual flood hours in the test period, the system captured 94.3% of all flood events at WATCH level or above, with only 5.7% receiving no advance warning. All code is developed in Python using open-source libraries and documented across three Jupyter notebooks structured around the CRISP-DM phases.

Date of Award

2026

Full Publication Date

2026

Access Rights

open access

Document Type

Undergraduate Project

Resource Type

bachelor thesis

Download

Share

COinS