| Abstract |
In car park occupancy problem, large amounts of data are collected from sensors and stored in databases. In order to discover useful information from such data, data mining techniques are applied. In this paper I propose to find alternative solutions for Birmingham car park occupancy issue. Our approach consist in clustering first the dataset in order to obtain relevant periods of time within a day and then forecast data within these clusters. Our experiments show that splitting data into six clusters and predict car park occupancy with k-Nearest Neighbor technique lead to the highest forecast rates. |
