A Supervised Learning Algorithm for Bike Sharing Demand Prediction
Abstract
Bike sharing is the new way of transport in coming years. Most of the developing countries are in the boom of this system. In India some of the entrepreneurs have tried their level best in order to implement this type of system in India and have failed to use data analytics correctly. There is a possibility that either the stations are completely full or empty when traveler come to station. With the help of our ML model organization can able to predict the number of bikes required on the hourly basis, so that they can maintain sufficient number of bikes and increase their profit with the help of our model. The data set is taken from the UCI repository. Machine Learning algorithm like Lasso, Linear Regression, Ridge, Huber Regressor, Decision tree regressor, Elastic Net CV, Gradient Boost, Random Forest Regressor, Extra Tree Regressor are used in order to predict the total number of bikes required at particular hour. After doing comparisons Random Forest Regressor gave the top accuracy score. So that top score given is used to for predicting the count of bikes on hourly basis.