Evaluating the performance of Random Forest, Decision Tree, Support Vector Regression and Gradient Boosting for streamflow prediction

This study utilized a range of machine learning algorithms to predict the hourly streamflow in the Ikpoba River. Data gathering relied on a Hydromet System installed along the river, collecting hourly measurements of gage height, ambient temperature, and atmospheric pressure. To convert the gage he...

Full description

Saved in:
Bibliographic Details
Main Authors: Osahon Idemudia, Jacob Odeh Ehiorobo, Christopher Osadolor Izinyon, Idowu Ilaboya
Format: Article
Language:English
Published: Can Tho University Publisher 2024-07-01
Series:CTU Journal of Innovation and Sustainable Development
Subjects:
Online Access:https://ctujs.ctu.edu.vn/index.php/ctujs/article/view/827
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849716906519429120
author Osahon Idemudia
Jacob Odeh Ehiorobo
Christopher Osadolor Izinyon
Idowu Ilaboya
author_facet Osahon Idemudia
Jacob Odeh Ehiorobo
Christopher Osadolor Izinyon
Idowu Ilaboya
author_sort Osahon Idemudia
collection DOAJ
description This study utilized a range of machine learning algorithms to predict the hourly streamflow in the Ikpoba River. Data gathering relied on a Hydromet System installed along the river, collecting hourly measurements of gage height, ambient temperature, and atmospheric pressure. To convert the gage height to streamflow data, historical gage and streamflow data covering the period from 2015 to 2020 were extracted from the Ikpoba River rating curve and were analyzed using curve fitting techniques to establish the precise relationship between streamflow and gage height. Various goodness-of-fit measures, such as adjusted R-squared value, standard error of estimate, and coefficient of determination, were utilized to identify the best-fit relationship. The estimated streamflow data were subsequently validated using the Soil and Water Assessment Tool, incorporating the digital elevation model of the study area, along with other input parameters like soil, slope, daily maximum precipitation, and daily maximum temperature. Validation results were illustrated using regression plots generated in Microsoft Excel. From the machine learning results, random forest algorithm outperformed other methods in predicting streamflow, with a mean square error of 0.02 and a coefficient of determination of 0.98. Conversely, decision trees showed superior accuracy in predicting individual data points, with the lowest root-mean-square error of 0.02.
format Article
id doaj-art-9c2fb5c450ea448cb05fb6ec16a72268
institution DOAJ
issn 2588-1418
2815-6412
language English
publishDate 2024-07-01
publisher Can Tho University Publisher
record_format Article
series CTU Journal of Innovation and Sustainable Development
spelling doaj-art-9c2fb5c450ea448cb05fb6ec16a722682025-08-20T03:12:50ZengCan Tho University PublisherCTU Journal of Innovation and Sustainable Development2588-14182815-64122024-07-0116210.22144/ctujoisd.2024.297Evaluating the performance of Random Forest, Decision Tree, Support Vector Regression and Gradient Boosting for streamflow predictionOsahon Idemudia0Jacob Odeh Ehiorobo1Christopher Osadolor Izinyon2Idowu Ilaboya3University of BeninUniversity of BeninUniversity of Benina:1:{s:5:"en_US";s:19:"University of Benin";} This study utilized a range of machine learning algorithms to predict the hourly streamflow in the Ikpoba River. Data gathering relied on a Hydromet System installed along the river, collecting hourly measurements of gage height, ambient temperature, and atmospheric pressure. To convert the gage height to streamflow data, historical gage and streamflow data covering the period from 2015 to 2020 were extracted from the Ikpoba River rating curve and were analyzed using curve fitting techniques to establish the precise relationship between streamflow and gage height. Various goodness-of-fit measures, such as adjusted R-squared value, standard error of estimate, and coefficient of determination, were utilized to identify the best-fit relationship. The estimated streamflow data were subsequently validated using the Soil and Water Assessment Tool, incorporating the digital elevation model of the study area, along with other input parameters like soil, slope, daily maximum precipitation, and daily maximum temperature. Validation results were illustrated using regression plots generated in Microsoft Excel. From the machine learning results, random forest algorithm outperformed other methods in predicting streamflow, with a mean square error of 0.02 and a coefficient of determination of 0.98. Conversely, decision trees showed superior accuracy in predicting individual data points, with the lowest root-mean-square error of 0.02. https://ctujs.ctu.edu.vn/index.php/ctujs/article/view/827Machine Learning, Random Forest (RF), Decision Tress (DT), Support Vector Regression (SVR), and Gradient Boosting (GB)
spellingShingle Osahon Idemudia
Jacob Odeh Ehiorobo
Christopher Osadolor Izinyon
Idowu Ilaboya
Evaluating the performance of Random Forest, Decision Tree, Support Vector Regression and Gradient Boosting for streamflow prediction
CTU Journal of Innovation and Sustainable Development
Machine Learning, Random Forest (RF), Decision Tress (DT), Support Vector Regression (SVR), and Gradient Boosting (GB)
title Evaluating the performance of Random Forest, Decision Tree, Support Vector Regression and Gradient Boosting for streamflow prediction
title_full Evaluating the performance of Random Forest, Decision Tree, Support Vector Regression and Gradient Boosting for streamflow prediction
title_fullStr Evaluating the performance of Random Forest, Decision Tree, Support Vector Regression and Gradient Boosting for streamflow prediction
title_full_unstemmed Evaluating the performance of Random Forest, Decision Tree, Support Vector Regression and Gradient Boosting for streamflow prediction
title_short Evaluating the performance of Random Forest, Decision Tree, Support Vector Regression and Gradient Boosting for streamflow prediction
title_sort evaluating the performance of random forest decision tree support vector regression and gradient boosting for streamflow prediction
topic Machine Learning, Random Forest (RF), Decision Tress (DT), Support Vector Regression (SVR), and Gradient Boosting (GB)
url https://ctujs.ctu.edu.vn/index.php/ctujs/article/view/827
work_keys_str_mv AT osahonidemudia evaluatingtheperformanceofrandomforestdecisiontreesupportvectorregressionandgradientboostingforstreamflowprediction
AT jacobodehehiorobo evaluatingtheperformanceofrandomforestdecisiontreesupportvectorregressionandgradientboostingforstreamflowprediction
AT christopherosadolorizinyon evaluatingtheperformanceofrandomforestdecisiontreesupportvectorregressionandgradientboostingforstreamflowprediction
AT idowuilaboya evaluatingtheperformanceofrandomforestdecisiontreesupportvectorregressionandgradientboostingforstreamflowprediction