Path Loss Prediction Accuracy Based On Random Forest Algorithm in Palembang City Area

Path loss is a mechanism where the signal from transmitting antenna to receiver in wireless network attenuated during transmission across medium due external field conditions. In telecommunication design, precise and efficient calculations are required. Random forest, as machine learning-based path prediction model, used this study. Machine prediction, random has low level of complexity high predictability. The data was collected using drive test method at Trans Musi busway area on 4G Palembang, South Sumatra, Indonesia. ratio comprised 20% testing set rest training set. As result, it obtained that accuracy 9.24% mean absolute percentage error (MAPE) root square (RMSE) 13.6 decibels (dB). Using hyperparameter tuning for forest results optimizing model used, resulting 8.00% MAPE RMSE 11.8 dB, which better than previous results.