Where can I get my electrical car charged in germany?

Everyone is talking about electro mobility — this caught my interest and I wanted to know more about the charging station infrastructure in Germany.

On the way, there are exiting ways to tackle data problems analytically as well as fast geographic visualization tecniques. I am going to explain the used packages and provide links to suitable documentations. In between the written words there will be python code elements to support the analysis. So let’s get into it.

The data-object which was returned consists of meta data and a key called elements where the desired data was stored. For initial exploratoy data analysis pythons panda package is my prefered tool when it comes to tabular data. Therefore we use a json normalizer to shrink dimensionality a put the data into a pandas data frame.

In october 2022 there appear about 19.5K rows and 682 columns. That means about 19.5K charging stations are listed in this data set — powered by this awesome community this is increasing every day. In contrast the amount of columns is also increasing since there is a freedom in creating new features andthe way of filling data in is somehow not very clear in a way. Therefore about 98% of all cells are just empty — missing values.

For this purpose the columns id, lat, lon and the operator are apparently most interesting storing the individual geographical position of the charging station in germany and the operator.

After mastering this hurdle, one could use DBSCAN algorithm to tackle to questions:
1. Are there stations where there is no other station in the range of 10 kilometers?
2. Where is the highest density of charging stations?

Therefore DBSCAN acts like a clustering algorithm without at forehand saying how many clusters to expect or indicating a certain form of the cluster. Choosing eps too large, the whole dataset is just one cluster. Choosing eps too small will result in singular points and just noise.

From a plain data viewpoint this DBSCAN-algorithm splitted the dataset in three categories where now two are visible in a geographical context right now — the dense stations, the lonesome stations and the rest — mutually exclusive. Embedding this analysis in a context of density where people live in germany, then this map is no surprise at all. One could go further here and add zip codes geometries and experiment with population density as well. From a GIS-technique viewpoint this would lead to more complex routines where packages like geopandas or rtree come into play. For now I leave it as is and focus on operators and local clusters.

Taking a closer look at column tags.operator shows that some are not homogeneously written. Therefore a brief cleansing is performed in the next code part. This should be adopted whenever there is new data since other anomalies could appear.

Basically, there will be two different operations in the next code block. First, creating a layer for every operator from the above list and then the rest of them. In every loop tons of folium circle objects are created and added to the map with corresponding coordinates and colors. When clicking on the different operators you will find that they are either very local players or equally distributed over whole germany.

Going deeper into regional operators we create a mapping to figure out which operators are regional and which are not. For the sake of simplicity this has been hard encoded via a dictionary.

This gives a glance at why geographical maps are very rarely used in business intelligence visualizations — it can come along pretty confusing and not too informative. But here is a little wrap up of what has been found in this analysis:

Charging stations are not equally distributed over germany. In metropolitan areas there is a higher availability than in rural areas. This is somehow a validation of what one might think when considering data of charging stations. When looking carefully at operator data and clustering there exist regional operators and non-regional operators. Regional operators are usually municipal suppliers who are driving the electromobility change on a local basis.

Then, there are retailers who use their location infrastructure to attract drivers of e-vehicles like Aldi, Lidl, Ikea or Kaufland, all very much equally distributed. Different to that are energy operators like Innogy, EnBW or RWE to name just a few have their core area from where they seem to have spread all over germany. In contrast to these, there seem to be Tesla or eon with equal distribution as a strategy.

Related posts:

Businesses worldwide leverage SaaS products in their everyday operations, especially with the rise of the remote work culture. From sales to accounting, SaaS platforms make it easier for companies to…

Here is a quick and easy way to prepare Pudina rice at home. It will hardly take 30 minutes including the boiling of rice. Before we jump on this delicious recipe let us first learn some benefits…

Every now and then we wish to tweak some of the features of WooCommerce. If you are selling variable products in your store, then you might have noticed the price range for the given product(s)…