Scholarlink

The change in weather patterns, particularly in rainfall intensity and distribution, is likely to disrupt the urban drainage system, thereby increasing the risk of flooding in vulnerable areas. This paper presents a comparative discussion of three machine learning models: Neural Networks (NN), Random Forest (RF), and Decision Trees (DT), to develop a model for predicting floods in Kuala Krai town, located in the central part of Kelantan, Malaysia. Six important characteristics in terms of meteorological and hydrological variables (water level in cm, monthly rainfall in mm, daily rainfall in mm, humidity in percent, temperature in degrees Celsius, and wind speed in m/s) were used to train and test the predictive models. It was found that the Neural Network model demonstrated the highest accuracy on average (98.9%), compared to the Decision Tree model (97.8%) and the Random Forest model (95.6%). In addition, the NN model has proved to be better in precision and F1-score, which tells us that it is more reliable in predicting any floods in the study area.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Evaluating Flood Prediction Algorithms Through a Comparative Analysis of Data Mining Techniques

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Mohd Zaki Mohd Salikon,
  • Aida Mustapha,
  • Mustafa M. Aljumaily,
  • Salam Omar Alo

摘要

The change in weather patterns, particularly in rainfall intensity and distribution, is likely to disrupt the urban drainage system, thereby increasing the risk of flooding in vulnerable areas. This paper presents a comparative discussion of three machine learning models: Neural Networks (NN), Random Forest (RF), and Decision Trees (DT), to develop a model for predicting floods in Kuala Krai town, located in the central part of Kelantan, Malaysia. Six important characteristics in terms of meteorological and hydrological variables (water level in cm, monthly rainfall in mm, daily rainfall in mm, humidity in percent, temperature in degrees Celsius, and wind speed in m/s) were used to train and test the predictive models. It was found that the Neural Network model demonstrated the highest accuracy on average (98.9%), compared to the Decision Tree model (97.8%) and the Random Forest model (95.6%). In addition, the NN model has proved to be better in precision and F1-score, which tells us that it is more reliable in predicting any floods in the study area.