Comparative Study of Imputation Methods for Weather Forecasting Systems in the Adamawa Region, Cameroon

Machine learning-based weather forecasting models are of paramount importance for almost all sectors of human activity. However, incorrect weather forecasts can have serious consequences on a global scale. The Adamawa region of Cameroon has suffered the consequences of erroneous forecasts, resulting in enormous losses and material damage. Missing or abnormal values are one of the problems that can contribute to the inaccuracy of weather forecasting models. Our objective is to provide a systematic view of how anomalies can affect the results of weather forecasting models and to compare imputation methods such as LinearRegressor, BayesianRidge, ExtraTreeRegressor (ETR), KNeighborsRegressor (KNR), DecisionTreeRegressor, KNNImputer, and MICE. The performance criteria used to evaluate these methods are RMSE and execution time. For the dataset, we generated four datasets with a specific anomaly rate. The abnormal values were transformed into missing values. Next, all the missing values were imputed using these different imputation methods, and then compared. Finally, the cleaned data was used by Machine learning-based forecasting models to generate forecast data. The results show that the imputation of anomalies in weather data is done in a reduced time, allowing for good quality weather data while improving the accuracy of weather forecasting models.