Irish Electric Vehicle Charge Point Status Datasets

I recently completed a minor thesis as a partial requirement for an MSc in computer science, a course with heavy leanings towards machine learning and data analytics.  The thesis explored the question of whether predictive analytics can be used to predict electric vehicle (EV) charge point availability in Ireland.

Contention for charge points is of increasing concern to Irish EV owners as the ratio of plugin EVs to charge points is set to rapidly increase in the near future, despite there being no plans to increase investment in the infrastructure.  A key motivation for the research was the idea that an algorithm that can make better-than-chance predictions about the availability and reliability of charge points from historical data, can potentially be used to inform a vehicle routing algorithm of charging stations to avoid when making route decisions for electric vehicles.

Unfortunately there were no datasets available to us lowly MSc students, so a big part of the workload involved building my own from live charge point status data provided in Ireland by ESB E-Cars, the organisation which at the time of writing, is responsible for maintaining the publicly funded charge-point networks in the Republic of Ireland and Northern Ireland. The live status data is available on the E-Cars charge point map and E-Car Connect app.

While the thesis only considered data from November 2016 to June 2017, I have continued to collect data and have made it accessible to EV drivers through a web site (  Although it is a work in progress and currently has limited search capabilities, CPInfo has proven to be a useful tool in evaluating the reliability and potential availability of charge points.  EV drivers use this information when planning their routes as it can help to identify charge points that are frequently out of service or occupied by other drivers.

Open Data Licence

I am now making the raw datasets available for anyone to use under a creative commons attribution 4.0 international public licence.  Licencing is necessary because considerable pre-prossing has been conducted on the source data to produce the datasets and this work attracts copyright protection implicitly. Explicitly offering an open data licence provides clarity on how the data can be used. Essentially, the licence allows anyone to use the data for any purpose on the condition that the source is correctly attributed. To attribute the source of the datasets, simply provide a hyperlink or citation referencing this blog post.

The licencing of the formatted datasets does not undermine the copyright that may be held by ESB E-Cars, who might own aspects of the datasets by virtue of owning the source data.  A representative of ESB E-cars has described this data as “publicly available and free to use”.

Dataset Details

The datasets take the form of monthly tab-delimited text files. Each line includes the date, time, charge point Id, charge point type {StandardType2, CHAdeMO, CCS, FastAC}, latitudinal and logitudinal coordinates, status {OOS, OOC, Part, Occ, Unknown} and address of a single charge point.  The data represent a snapshot of the status of the charge point network taken at five-minute intervals.

StandardType2 is a slow charge point of up to 22kw AC, while the CHAdeMO, CCS and FastAC are different types of DC and AC rapid charge points.  A standard type two charge point represented in the datasets typically has two available connections that can be used simultaneously, while a rapid charge point can have either one, two or three connections, each with a different connection type (CHAdeMO, CCS or FastAC).

The available status has been omitted as it can be implied by the absence of a record. The other statuses include out of service (OOS), out of contact (OOC), partially occupied (Part), fully occupied (Occ) and Unknown.  An unknown status exists where the status data was either not available or otherwise not polled due to connection issues at the time interval in question.  Where the status is unknown for all charge points at the interval, a single record exists with a charge point id of unknown and a status of unknown.

There are also a number of nuances in the data that must be considered.

First of all, charge points are sometimes moved from one location to another and/or replaced by another charge point. When this happens, the charge point Id at a given location changes in the dataset.  Furthermore, the charge point removed from one location, can appear again at another. The charge point Id therefore cannot be relied upon as a means to track the charging activity at a particular location. Instead, the latitudinal and longitudinal coordinates should be used. However, these can also change slightly (by a matter of meters) and thus can’t be used directly as unique identifiers. To get the full history of activity at a given location, a search should include charge points within a tight range of latitudinal and longitudinal values rather than by matching the exact values.

Another nuance is that the charge point map sometimes goes through periods, of up to several days, where the status data are not updated. This is reflected in the datasets by statuses that don’t change for unusually long periods of time.  As it stands, the datasets reflect the state of the charge point maps during those periods rather than the statuses of the charge points in reality.  These periods can be identified via a checksum calculated over all rows at each time interval or by manually checking the statuses at the busiest charge points.

Rapid chargers never hold a status of partly occupied despite the fact that in many cases it is possible to charge on the FastAC connection at the same time as CCS or CHAdeMO connection.  Furthermore, an occupied status on the CCS or CHAdeMO connection does not necessarily imply that that specific connection was in use.  This is because only one of the two can be used at any given time and thus if the CHAdeMO connection is in use, then CCS is also unavailable and vice versa.

At the time of writing there are no examples of multiple rapid charge points in the same vicinity, however, there are a small number of cases where there are multiple standard type two charge points at the same location.  Dundrum town centre and the Stillorgan Luas station are two examples.  On the map, these are represented slightly differently to other charge points in that the charge point Ids and statuses are grouped together.  If there are two charge points, and therefore four connections, only one icon appears on the map listing both charge point ids and the status will only show as fully occupied if all four connections are in use.  Consequently, these were omitted from the dataset.  I intend to rectify this in a future script update.

Future Changes and Download Links

Any updates to the datasets will be expressed by updating this blog post.  The datasets are available for download here, where they will be updated monthly.

Author: James Burkill

Software engineer and student of all things AI. LinkedIn:

Leave a Reply