Application of Long Short Term Memory (LSTM) and Gated Recurrent Unit (GRU) Models to Determine Prices on Ride Hailing

Abstract

Ride hailing is a transportation system that is growing rapidly to date. One of the factors that influence people to use ride-hailing is related to travel costs. The cost of the trip is determined by several indicators. But by using machine learning the required indicators are less, compared to using conventional methods. When creates a machine learning requires hyperparameters, as a machine learning framework, and data. The data will be processed so that it can be used to create a machine learning. The hyperparameters that will be considered are the model to be used, the size of the epoch, the proportion of data sharing, and the application of Min Max Scalar normalization. There are 2 types of models that will be the basis for machine learning, namely LSTM and GRU. When determining a good combination of engine hyperparameters, the 10 best engines will be selected. The best result of the trial program from this research is the program with the LSTM model with 500 epochs added with the use of MMS which divides the existing data by the proportion of 0.25 with loss 0.7669816613197327. The main conclusion obtained in this research is that the implementation of the model used is quite good, so the results obtained have a fairly satisfactory loss value.