Abstract:
Small and medium enterprises (SMEs) play major roles in achieving economic
development in countries all over the world. Currently, Sri Lanka doesn’t have
a generally accepted criterion to categorize SMEs. Different agencies use
various criteria and there is no consistency between them. The most common
criterion is the number of employees in the company. Though this is simple, it
neglects an important characteristic such as annual turnover. Hence, a
company with fewer employees and large turnover categories to small scale
enterprises. So, Identifying SMEs on a commonly acceptable criterion is a long
felt need of the country. Department of Census and Statistics, Sri Lanka
currently categorize SMEs based on number of employees. This classification
doesn’t consider the important variables such as industry turnover, capital and
energy consumption. The main focus of this study is to propose an alternative
method to categorize Small and Medium Enterprises in Sri Lanka. Data were
collected from Annual Survey of Industries (ASI) which is conducted by
Industries, Trade, Construction and Services Division of Department of Census
and Statistics. Number of Persons Engaged In, Annual Turnover of the
Industry, Annual Energy Consumption of the Industry and Fixed Capital Assets
were identified as the key variables to categorize Industries. In our approach
Industries are clustered based on the above four variables using Model Based
Clustering (EM) Algorithm. Then a Decision Tree algorithm is used to define
the cluster boundaries for small, medium and large industries. Finally, a rule
set containing nine rules was prepared and proposed to classify industries. The
proposed method has the ability to capture the multi-dimensional behavior and
hence, produce more accurate categorization.
