Arranging Bus Behaviour by Finding the Best Prediction Model with Artificial Neural Networks

Abstract

Artificial Neural Networks (ANNs) were used in this study to estimate the hourly passenger populations at certain stations in İstanbul. To do this, the details were collected from various sources regarding the passengers in a station. This study aims to show what can be implemented for the passenger numbers in the decision support system and makes some recommendations for the regulation of the bus lines. Trials were conducted using an ANN with a backpropagation model and various inner layers for the estimations. The MAE score was 10.301 for the stations studied. Qualitative interviews were conducted with 32 passengers and 12 drivers, and solutions were searched for the density of the lines. A proposal system was developed with the c# software resulting from the combination of the prediction model with these proposals.

LSTM Deep Neural Network Based Power Data Credit Tagging Technology

“…To sum up, based on the above research results, scholars continue to innovate and improve enterprise risk assessment, in particular, the application of neural network algorithms is advantageous [18][19][20]. Modern credit risk management techniques have become more sophisticated.…”

A K-means Group Division and LSTM Based Method for Hotel Demand Forecasting

“…In addition, the construction of the model is complex and requires high statistical expertise. In recent years, machine learning and deep learning have been proven to have good demand forecasting capabilities and can effectively capture nonlinear and complex features in time series [16][17][18][19], but they have not been widely used in the hospitality. Given the reservation and occupancy history records, ridge regression, kernel ridge regression, multilayer perceptron, and radial basis function networks are constructed to forecast the daily occupancy rate, and the good forecasting performance is obtained [20]; Aliyevetc.…”