Property Business Classification Model Based on Indonesia E-Commerce Data

Abstract

Online property business or known as e-commerce is currently experiencing an increase of home sales.
Indonesia e-commerce property business has positive trending shown by the increasing sales for more than
500% from 2011 to 2015. A prediction of property price is important to help investor or public to have accurate
information before buying property. One of methods for prediction is classification based on several distinctive
of property industry attributes, such as building size, land size, number of rooms, and location.
Today, data is easily obtained, there are many open data from E-commerce sites. The E-commerce contains
information about home and other property advertised to sell. People also regularly visit the site to find the
right property or to sell the property using price information which collectively available as open data.
To predict the property sales, this research employed two different classification methods in Data Mining which
are Decision Tree and k-NN classification. We compare which model classification is better to predict property
price and their attributes. We use Indonesia biggest property based ecommerce sites Rumah123.com as our
open data source, and choose location Bandung in our experiment. The accuracy result of decision tree is 75%
and KNN is 71%, other than that k-NN can explore more data pattern than Decision Tree.