Create a Sample Set of PLZ for Clustering

Motivation

To capture the diversity of different types of settlements and represent them for the chosen region adequately a sampling algorithm is developed. For the area of interest that could be a Bundesland or Germany generating grids for the whole region takes to long. The number of PLZ is reduced from 8181 (Germany) or e.g. 2065 (Bayern) to 100 PLZ in the representative sample set.

Usage

The sample set is created with the function:

The result can be retrieved from the database as a DataFrame with

Joining Regiostar with the Municipal Register

In order to sample areas from the Regiostar Classes they need to be mapped with the Municipal Register (Gemeindeverzeichnis) from Germany. You can refer to ../../docs_pylovo/municipal_register/municipal_register for more details.

Sampling Algorithm

To make sure all settlement types are represented in the sample set the Regio 7 Dataset with seven different types of municipalities is chosen.

The share of PLZ samples from each RegioStar 7 class corresponds to the population share of the class of the total population. (An analysis by the Chair of Renewable and Sustainable Energy Systems at TUM have shown that the percentage of energy usage of a Regio7 class corresponds to its population share.)

For Bayern the distribution of samples are:

Within the classes the population density distribution of the class is reproduced.