Forecasting Flood in Bangladesh Using Different Time Series Models: A Comparative Analysis

Natural disasters frequently strike Bangladesh. Floods and cyclones are frequent natural disasters among them. These hazards occur in Bangladesh at certain times and harm millions of people’s lives and their belongings and bring on a severe economic crisis for a developing nation. To mitigate these losses, we observed and decided to understand the reasons behind it and tried to find a solution. We decided to gather data on three factors that can directly impact weather and natural hazards. Temperature, rainfall, and rivers water-level are the factors as these three are correlated. Since finding a solution is difficult, we gathered historical data on these parameters and used a machine-learning approach. Time series forecasting works well with historical data therefore, this method can bring some outcomes that can result to forecast some reflection that can help to diminish these constant calamities. This paper demonstrates the implementation and forecasting of time series models employing (ARIMA, SARIMAX, and PROPHET) for future forecasts to decrease catastrophes in Bangladesh.