Imputation methods for missing values: the case of Senegalese meteorological data

Abstract

nge studies require comprehensive databases to analyze the climate signal, to monitor its evolution, and to predict more accurately future changes. Since complete observations of any continuous process is almost impossible, it is then inevitable to encounter missing information in meteorological databases. The aim of this work is to evaluate the performance of five ($5$) imputation methods: missForest, $k$-nn, ppca, mice and imputeTS. The results show that missForest is the best performing method to handle missing temperature data. In the case of precipitation data, the imputeTS method is the preferred one.