Heating Energy Performance Assessment using KNN Classification method: Investigating Test Size, Neighbor Parameter, and Class Count Effects
Аннотация
We analyze the energy requirements for assessing the heating load of building systems using the k-Nearest Neighbor (KNN) method for classification. Unsupervised transformations of the dataset are employed in learning algorithms. Two distinct data representation classes are generated separately from the original dataset based on captured heating load magnitudes. The unsupervised transformations of the dataset yield 4 and 8 classes, respectively.To evaluate performance, various test and training sizes are employed, comparing measurement outcomes and predicting class label descriptions for two dataset representations as the parameter for the nearest neighbor (k) ranges from 1 to 20. Results obtained from the simulation indicate that accuracy is influenced by factors such as test size, the k parameter, and dataset class count. The training and test accuracy rates reach maximum values of 95% and 85%, respectively. Keeping k constant, the test size has a negligible impact on accuracy. The classes exhibit three critical regions: 1) Region I, where accuracy fluctuates with increasing k in the interval [1]-[5], and 2) Region II, where accuracy declines with increasing k in the interval [5]-[10]. In Region III k [10]-[20], accuracy remains relatively constant. While reducing the number of classes may result in a slight improvement in accuracy by simplifying the classification task, the trade-offs and complexities of increasing the number of classes for data representation should be carefully considered. Although this approach can lead to a more flexible design, it may slightly reduce accuracy.