Create Spatial Weight Matrix for Provinces in Java Island

The spatial weight matrix is one crucial component in the spatial analysis models as it represents the relationship between one region with others.

There are two kinds of spatial weighting matrix: contiguity and distance-based matrix. In a contiguity matrix, two regions are considered as neighbour if they share the same border, that represented with 1 symbol. While not neighbouring regions will be coded 0. On the other hand, each cell in the distance-based matrix shows the distance between the two regions.

 

Figure 1. Java Island, Indonesia

In this article, we will create a spatial weight matrix of provinces in Java Island (Figure 1). For the six provinces in Java Island, there will be a 6x6 matrix.

The creation of a contiguity matrix should be started by identifying the relationship status between one region and others (Table 1). Then, the contiguity matrix is created (Table 2)

Table 1. The Relationship between Regions

Table 2. Contiguity Matrix

One drawback of the application of the contiguity matrix is it cannot deal with water/sea border. If using a contiguity matrix. Banten has a relationship only with DKI Jakarta and Jawa Barat but does not have a relationship with Lampung Province as they are separated by sea. Similarly, if using a contiguity matrix, Jawa Timur only has neighbour of Jawa Tengah but it is not a neighbour for Bali as they have a water area as a border.

To deal with this condition, the distance-based matrix is used. The distance is obtained from the centroids, i.e. is the central point of an area or a geographical centre. In figure 2,the centroids are showed by the dots.

 

Figure 2. Centroids in Java Island

Using Drukker, et.al. (2013) estimator in Stata , we can generate the inverse-distance matrix (Table 4). The summary of inverse distance matrix is presented in Table 3

 Table 3. Summary of Inverse Distance Matrix

Table 5 shows the distance of provinces in Java Island, measured from their centroids. The shortest distance of 75.66 km is between DI Yogyakarta and Jawa Tengah while the longest distance of 744.36 km is between Jawa Timur and Banten.

As the maximum distance between provinces in Java is 744.3799 km (inverse of 0.001343), the average distance among them are 269.3022 km (inverse of 0.003713), and the minimum distance is 75.66185 km (inverse of 0.013217).

Table 4. Inverse-distance Matrix

Table 5. Distance between Provinces in Java Island

Table 5 shows the distance of provinces in Java Island, measured from their centroids. The shortest distance of 75.66 km is between DI Yogyakarta and Jawa Tengah while the longest distance of 744.36 km is between Jawa Timur and Banten.

To sum up, two types of spatial weighting matrix can be used to explain the link between regions. This article provides a simple example of the application of a contiguity matrix and a distance-based matrix.

References:

Drukker, D. M., Peng, H., Prucha, I., & Raciborski, R. (2013). Creating and managing spatial-weighting matrices with the spmat command. Stata Journal, 13(2), 242-286

OCHA. 2019. Indonesia - Subnational Administrative Boundaries.

https://data.humdata.org/dataset/indonesia-administrative-boundary-polygons-lines-and-places-levels-0-4b

 

Popular posts from this blog

Skewness dan Kurtosis

Piramida Distribusi Kekayaan Masyarakat Indonesia

KKN di Desa Penari

Palma Ratio Indonesia

Daya Beli Masyarakat, in this Economy: Dunia Usaha dan Perspektif Ekonomi Makro

Generasi Hutang: Literasi Keuangan dan Kekayaan Rumah Tangga

Berapa Rata-Rata Kekayaan Rumah Tangga di Indonesia?