Data exploration in GRASS GIS – boxplots

I am currently working on some exercises for which I need data about municipalities in the Netherlands. A good place to look for such data is the CBS (Dutch Central Bureau of Statistics). One data layer is vector layers of the dutch municipalities and neighborhoods, which include demographic data.

One of the first things I normally do when exploring new data is to look at the distribution of the data. For example by creating a histogram using the d.vect.colhist addon (see my earlier post). But what if I want to compare the distribution of different groups or samples? In such a case I find boxplots more convenient. However, there is no tool in GRASS GIS to create boxplots, so I had a look at the d.vect.colhist addon code and adapted the code to create boxplots instead of histograms.

An example

Let’s for example look at the average population densities of the municipalities.

The average population density (number of inhabitants / km2) per municipality in 2017. Source: CBS.

What if I want to compare the distribution of the average population density per provinces Dutch provinces? You can install the addon (see the end of this post) and run d.vect.colbp on the command line or the console. This will open a window with different tabs.

In the first tab, you can define a column in the attribute table to plot (here BEV_DICHTH, which is the column with the population density) and a column that will be used to group the data (here provincie, which gives the names of the provinces the municipality belongs to). As you can see in the screenshot above, you have a few options to change the plot (layout). In this case, I choose to rotate the x-axis labels so they do not overlap. The resulting plot looks like:

The distribution of the average population densities of the Dutch municipalities per province.

You can of course also use the command line. In this case I will plot the boxplots horizontally using the ‘h flag’.

d.vect.colbp -h map=gemeenten@CBS column=BEV_DICHTH \
    where="AANT_INW > 1" plot_output=example_1.png \
    group_by=provincie order=ascending --overwrite

With will give you the plot below.

The distribution of the average population densities of the Dutch municipalities per province.

The add-on does not provide further options to change the appearance of the plot, as the main idea is to use this for quick exploration of your data, similar to the other plotting tools in GRASS GIS. However, you can save the plot as a svg file, and further edit it in e.g., Inkscape.

Testing the add-on

I probably will try to integrate the option to create boxplots in the existing d.vect.colhist add-on, but for now it is available as a separate addon on Github for testing. For GRASS GIS version 7.2 and above, you can use g.extension; simply type on the command line:

g.extension d.vect.colbp \
    url=https://github.com/ecodiv/d.vect.colbp

I am sure there is a lot to improve as this was a rather quick hack, so any feedback will be most welcome. If you try it out and run into problems, please let me know (suggestions for improvements are of course also welcome).

Advertisements

Draw a histogram of vector attribute column in GRASS GIS

GRASS GIS has convenient tools to draw histograms of raster values. As similar tool to draw a histogram of values in a vector attribute table lacks. But you can easily add this functionality by installing the d.vect.colhist addon by Moritz Lennert. Read this short post on Ecodiv.earth tutorials.

bitmap

Hands-on course to GIS and Remote Sensing with GRASS GIS

The hands-on GRASS GIS course at ITC – University of Twente on November 3rd, 2017 was a great success. The course, organized by ITC and OSGeo.nl, offered a very nice introduction to GRASS GIS by Veronica Andreo and a guided tour about working with GRASS GIS by Sajid Pareeth.

KLdistribution

As part of the course, we also developed three modules with hands-on exercises on different topics related to raster time series processing, remote sensing images processing and spatial interpolation in GRASS GIS.

All the course materials are available online, so check them out and enjoy 🙂

 

 

K-fold cross validation in GRASS GIS

A common technique to estimate the accuracy of a predictive model is k-fold cross-validation. In k-fold cross-validation, the original sample is randomly partitioned into a number of sub-samples with an approximately equal number of records. Of these sub-samples, a single sub-sample is retained as the validation data for testing the model, and the remaining sub-samples are combined to be used as training data. The cross-validation process is then repeated as many times as there are sub-samples, with each of the sub-samples used exactly once as the validation data (Table 1).

table_1
Table 1. Illustration of data partitioning in a 4-fold cross-validation, with training data used to train the model, and test data to validate the model.

The k evaluation results can then be averaged (or otherwise combined) to produce a single estimation. The advantage of this method is that all observations are used for both training and validation, and each observation is used for validation exactly once.

Functions for modelling and machine learning in e.g., R and Python’s Scikit-learn often contain build-in cross-validation routines. But it is also fairly easy to build such a routine yourself. This tutorial shows how one can easily build a k-fold cross-validation routine in GRASS GIS, e.g., to evaluate the predictive performance of two interpolation techniques, the inverse Distance Weighting and bilinear spline interpolation.

elevation_v1
Figure 1. A) Elevation map of North Carolina. B) Elevation estimation based on inverse distance weighting interpolation of the elevation at 150 random sample points. C) Residue map with the differences between A and B. D) Relative differences between A and B, computed as (A-B)/A. Map C and D are overlaid with the 150 sample locations.

This tutorial is available on https://tutorials.ecodiv.earth.

Plotting GRASS data in Python

GRASS GIS offers some useful but basic plotting options for raster data. However, for plotting of data in attribute tables and for more advanced graphs, we need to use other software tools. In this tutorial I explore some of the possibilities offered by Pandas plot() and how we can further tune plots using matplotlib / pyplot library.

map_municipals
Map of the municipals in Wake County, North Carolina, and for each municipal the distribution of distances to the nearest school (data source: North Carolina sample data set).