Outlier Detection in Indoor Localization Using KNN and Random Forest Classifier

Now a days, indoor localization is very much essential in wide range of applications like industry, shopping malls, and warehouses. So, this work is aimed to comparing the prediction of information from open data sets using K-Nearest Neighbor Classifier and Random Forest by analyzing the performance parameters of accuracy, and precision. Group 1 namely K-Nearest Neighbor and group 2 namely Random Forest were taken for indoor localization and 200 samples of each group were taken with pretest power of 80% for statistical analysis. KNN Classifier achieves an accuracy of 94.23% and precision of 85.1%. Random Forest achieves an accuracy of 87.23% and precision of 80.26%. From the statistical analysis, the significance was obtained below the 0.05 for both accuracy and precision. KNN Classifier achieves significantly better accuracy and precision when compared with RF Classifier.