URBAN DEVELOPMENT CLASSIFICATION OF AKWA-IBOM STATE, NIGERIA USING KMEANS ALGORITHM

¹Beulah Ofem, Onuwa Okwuashi², Mfon Isong², Okey Nwanekezie⁴, and Aniekan Eyoh²

¹Department of Urban and Regional Planning, University of Uyo, Uyo, Nigeria

^2,Department of Geoinformatics & Surveying, University of Uyo, Uyo, Nigeria

⁴Department of Estate Management, University of Uyo, Uyo, Nigeria

E-mail: onuwaokwuashi@yahoo.com,

ABSTRACT

K-means algorithm presents a less cumbersome technique for easy classification of urban centres based on empirical causal factors. The user of the algorithm has no influence over the classification result of the algorithm. Instead, the algorithm delineates urban centres by using cluster characteristics of the variables to define soft separation boundaries between or amongst k-classes. No training of the k-means algorithm is required since k-means is an unsupervised classifier. The k cluster centroid locations and sums of point-to-centroid distances are first computed, and thereafter the distances from each point to every centroid. The classification solution for each pixel is found by determining the class that yields the least computed distance from each point to every centroid; such that the successful class wins the classification for that pixel. This study presents the application of an unsupervised kmeans algorithm to the delineation of urban development centres in Akwa-Ibom State, Nigeria. Using 21 variables, 82 settlements drawn from 25 out of 31 local government areas of Akwa-Ibom State are classified into three levels of urban development: high, medium, and low. Eighty-two settlements experimented in this research are classified into eight high, twenty-two medium, and fifty-two low level development centres respectively.

Keywords: Classification; Kmeans; Urban Development