A Model for Predicting Flood Using Machine Learning Techniques: A Case of Adamawa State, Nigeria

A Model for Predicting Flood Using Machine Learning Techniques: A Case of Adamawa State, Nigeria

Hannatu G Mose¹, Afrifa James², Ayuba G John³, Ibrahim Robinson⁴

¹²⁴: Department of Computer Science, Adamawa State Polytechnic, Yola

³: Department of Disaster Management, Adamawa State Polytechnic, Yola

Keywords: Prediction, Machine Learning, Flood, Model Artificial Neural Network

Abstracts: In recent years, Adamawa State was badly affected by the 2022 flood due to heavy rainfall, climate change and the release of water from Lagdo Dam from neighbouring Cameroon. The 2022 flood is the worst affected flood in Nigeria’s history since 2012. This paper developed a model for predicting floods in Adamawa State North-East Nigeria using Artificial Neural Network. The study collected secondary data for the model. A total of 360 data samples spanning over twenty-eight (28) years comprising atmospheric and hydrological data were collected. The interval of data collected was between the years 1992 to 2021. The data collected consisted of six (8) initial columns. The data sample contains relevant parameters such as the year/month of the data, monthly mean of the max temperature, mean of the monthly minimum temperature, relative humidity, the volume of rainfall, gauge height and the class of the data. .  The collected data were pre-processed, cleaned, and modelled using Jupyter Notebook, a Python Anaconda development platform for Artificial Neural to build the predictive model. The model was evaluated using a confusion matrix. The prediction accuracy was 96.37% was obtained. The study concluded that Flood predictive model will alert individuals residing in the flooded area and recent authorities so that proper measures will be taken to curtail the problem of flooding in Adamawa State, Nigeria.