Hierarchical Regional Disparities and Potential Sector Identification Using Modified Agglomerative Clustering

1st Annual Applied Science and Engineering Conference

工业技术;自然科学

Munandar, T.A.^1,3 ; Azhari^1 ; Mushdholifah, A.^1 ; Arsyad, L.^2

Computer Sc. and Electronic Dept., Math. and Natural Science Faculty, Universitas Gadjah Mada, Yogyakarta, Indonesia^1

Economics Department, Faculty of Economics and Business, Universitas Gadjah Mada, Yogyakarta, Indonesia^2

Informatics Eng., Dept., Information Technology Faculty, Universitas Serang Raya, Indonesia^3

关键词: Agglomerative clustering;    Central Java Province;    Competitive advantage;    Hier-archical clustering;    Hierarchical agglomerative clustering;    Hierarchical clusters;    Regional development;    Regional disparities;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/180/1/012074/pdf
DOI  :  10.1088/1757-899X/180/1/012074

来源: IOP

【 摘 要 】

Disparities in regional development methods are commonly identified using the Klassen Typology and Location Quotient. Both methods typically use the data on the gross regional domestic product (GRDP) sectors of a particular region. The Klassen approach can identify regional disparities by classifying the GRDP sector data into four classes, namely Quadrants I, II, III, and IV. Each quadrant indicates a certain level of regional disparities based on the GRDP sector value of the said region. Meanwhile, the Location Quotient (LQ) is usually used to identify potential sectors in a particular region so as to determine which sectors are potential and which ones are not potential. LQ classifies each sector into three classes namely, the basic sector, the non-basic sector with a competitive advantage, and the non-basic sector which can only meet its own necessities. Both Klassen Typology and LQ are unable to visualize the relationship of achievements in the development clearly of each region and sector. This research aimed to develop a new approach to the identification of disparities in regional development in the form of hierarchical clustering. The method of Hierarchical Agglomerative Clustering (HAC) was employed as the basis of the hierarchical clustering model for identifying disparities in regional development. Modifications were made to HAC using the Klassen Typology and LQ. Then, HAC which had been modified using the Klassen Typology was called MHACK while HAC which had been modified using LQ was called MACLoQ. Both algorithms can be used to identify regional disparities (MHACK) and potential sectors (MACLoQ), respectively, in the form of hierarchical clusters. Based on the MHACK in 31 regencies in Central Java Province, it is identified that 3 regencies (Demak, Jepara, and Magelang City) fall into the category of developed and rapidly-growing regions, while the other 28 regencies fall into the category of developed but depressed regions. Results of the MACLoQ implementation suggest that there is only 1 regency which falls into the basic-sector category (Banyumas), while the other regencies fall into the non-basic non-competitive sector category.