January 8, 2014
Predicting the availability of parking spaces in Ljubljana car parks. Time series (well, time window) regression using linear regression, regression trees and random forest. Predicting for 30min, 1h, 2h and 3h intervals.
Car Parks are common and essential infrastructures in modern day cities. The availability of parking spaces can have an impact on a person’s choice of transportation mode, departure time or even on wether to depart at all. The city of Ljubljana provides the number of available spaces for each car park online. However, since decisions often need to be made ahead of time, the availably of parking spaces should be provided ahead of time. In this paper we analyse the viability of providing such information through the use of predictive modelling. We begin by describing how we built these models and conclude that they are both viable and accurate. Finally, we address how the models could be further improved.
- Machine learning part comes from Scikit-learn.
- Uses Pandas for data manipulation/transformation e.g. resampling.
- Plotting using Pandas / Pyplot
The data can be collected from Open Data Slovenia. Doubt I can share the dataset that was provided to me but you can always try collecting a new one.
The particular dataset that was provided to me consisted of a set of gzipped json files and required a lot of cleanup. I initially imported it into an sqlite database and did much of the cleanup there. The code for this is in the source code but I imagine it is next to useless for anyone else.
This was done as part of a course assignment for the Josef Stefan International Postgraduate School.
This is mostly an ugly version of a iPython Notebook since I was using tmux with vim and ipython panes... the code is uncommented and ugly but it is complementary to both the report and the presentation and might be useful if you plan on doing something similar.
I doubt I'll ever get around to cleaning up this mess but you never know.