Smote vs. Random Undersampling for Imbalanced Data - Car Ownership Demand Model

Because the numbers of cars reflect each person's travel behaviors for specific location, car ownership demand model plays a dominant role in analysis order to understand area's individual and household behaviors. However, study project master plan Khon Kaen expressway represented imbalanced data; namely, majority class minority were not equal. Before developing machine learning model, this suggested solution balance data by using oversampling under-sampling techniques. The data, which had been improved with SMOTE (Synthetic Minority Oversampling Technique) kNN (k-nearest neighbors) (k = 5), demonstrated better effect than other algorithms that studied. TPR (true positive rate) rural suburban areas, are types regions very different imbalance ratios, was calculated before balancing at 46.9 % 46.4 %. As result, values 63.5 54.4 %, respectively, following balancing.