Application of Data Mining in Pharmaceutical Imports in Kenya

Abstract

Data mining which is often referred to as Knowledge Discovery in Databases (KDD) is a subdiscipline of computer science aiming at the automatic interpretation of large datasets. Its end result is the extraction of meaningful patterns from data to aid in the process of decision making. Data mining is a discipline that is applicable to all subject areas from aviation to zoology as most have transaction data. Historical data from permits prior to the proposed system is of high value regardless and is of great interest to the study as it proposed to enhance usability of such data for decision purposes. All data on imports and exports available electronically was captured through the data capture system. Currently, PPB uses regular queries in its databases to generate reports. These reports only present one facet of data without presenting the true picture to the concerned decision makers leading to inaccurate decisions. Objectives of the study were to analyse data on imports and exports of pharmaceutical products in Kenya and discover patterns of association and correlation between the various pharmaceutical product groups. The study adopted the CRISP-DM framework for data mining and utilised RapidMiner as a tool for data analysis and mining. CRISP-DM is a generic framework that is applicable in most subject areas and is quite tested. RapidMiner is an open source data mining tool that is quite popular world-wide due to its capabilities, ease of use and availability of online help. This study applied data mining to the field of pharmacy regulation. The study analysed data on imports of pharmaceutical products for interesting patterns. The study performed correlation and association analysis on the import data. The study proved that there exist patterns in pharmaceutical import data. The patterns are also similar to prescription patterns from studies in Ethiopia, Nigeria and India especially on association of several pharmaceutical product groups.