NORMA eResearch @NCI Library

Efficiency of Flash Flood Prediction by XGBoost and Random Forest using 15 minutes & 1 hour time period sensor data.

Iyer, Ghiridhar (2020) Efficiency of Flash Flood Prediction by XGBoost and Random Forest using 15 minutes & 1 hour time period sensor data. Masters thesis, Dublin, National College of Ireland.

[thumbnail of Master of Science]
Preview
PDF (Master of Science)
Download (1MB) | Preview
[thumbnail of Configuration manual]
Preview
PDF (Configuration manual)
Download (5MB) | Preview

Abstract

Floods are one of the costliest and deadliest Natural Disasters known to mankind. Due to the inconsistent nature of rain, estimation of flood becomes complex. Most of the previous works have focused on forecasting floods but limited research has been done on flash flood prediction also known as nowcasting. Since Flash Floods manifest in a matter of hours, people remain unaware of the disaster leading to loss of lives. Many previous works have highlighted the time period (time difference between successive rows) of the dataset as the limitation to predict flash floods. By foreseeing the disaster as well as assessing its threat in real-time would ensure timely actions which can avoid loss of life. This paper predicts flash floods using XGBoost and Random Forest based on UK Sensor Data. This paper also examines the effect of the time period of the dataset on the performance of the prediction model. AWS Platform was used to host the application. GAN was utilised to mimic the dataset and increase the number of records. Algorithms were scripted and were provided to the Sagemaker ML endpoint for training and prediction. Both the algorithms successfully predicted flash floods and river level for about 3 days. The PASS evaluation technique has been adopted for assessing the performance of algorithms. XGBoost outperformed Random Forest in all evaluation aspects and hence saves time and lives of the people. Implementation and performance assessment of Neural Networks is yet to be performed.

Item Type: Thesis (Masters)
Subjects: Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Cloud computing
Divisions: School of Computing > Master of Science in Cloud Computing
Depositing User: Dan English
Date Deposited: 28 Jan 2021 14:16
Last Modified: 28 Jan 2021 14:16
URI: https://norma.ncirl.ie/id/eprint/4537

Actions (login required)

View Item View Item