The functional activities questionnaire and neuropsychological scores are important tools for assessing the cognitive and functional domains of Alzheimer’s disease patients. Additionally, standardized datasets that are curated from multiple global centers are currently available and help in the computer-aided diagnosis systems development. But there are a lot of clinical tests available to gauge these scores, which makes diagnosing them is a difficult task. Additionally, the datasets have widespread problems with missing and unbalanced data. To address these problems, we suggest a machine learning-based system in this research. Scientific findings show that the decision tree performs better on the neuropsychological score following the miss forest imputation and on the functional activities questionnaire scores following the application of the synthetic minority oversampling technique.