A machine learning approach in Python is used to forecast the number of train passengers using a fuzzy time series model

Solikhin Solikhin, Septia Lutfi, Purnomo Purnomo, Hardiwinoto Hardiwinoto

Abstract


Train passenger forecasting assists in planning, resource use, and system management. forecasts rail ridership. Train passenger predictions help prevent stranded passengers and empty seats. Simulating rail transport requires a low-error model. We developed a fuzzy time series forecasting model. Using historical data was the goal. This concept predicts future railway passengers using Holt's double exponential smoothing (DES) and a fuzzy time series technique based on a rate-of-change algorithm. Holt's DES predicts the next period using a fuzzy time series and the rate of change. This method improves prediction accuracy by using event discretization. positive, since changing dynamics reveal trends and seasonality. It uses event discretization and machine-learning-optimized frequency partitioning. The suggested method is compared to existing train passenger forecasting methods. This study has a low average forecasting error and a mean squared error.

Keywords


Frequency-based partitioning; Machine learning; Prediction; Rate of change; Transportation public

Full Text:

PDF


DOI: https://doi.org/10.11591/eei.v11i5.3518

Refbacks

  • There are currently no refbacks.


Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

Bulletin of EEI Stats