USAGE OF DATA MINING FOR EVALUATION OF BORSA İSTANBUL REGISTERED COMPANIES’ FINANCIAL STRUCTURE

Yıl 2017, Cilt: 6 Sayı: 1, 34 - 38, 30.12.2017

Mehmet Ozkan Levent Boran

https://doi.org/10.17261/Pressacademia.2017.742

Öz

Objective-
Financial analysis are mostly done for evaluation of companies’ financing
and investment needs with traditional analysis methods such as vertical
analysis, horizontal analysis and ratio analysis. Although these traditional
methods support the analyst for single company evaluation, they are inefficient
while questioning many companies. Therefore, decision makers face
time-consuming problem when they evaluate hundreds of companies, which are
necessary for profit maximization, cash flow maximization and risk mitigation
etc. It is aimed to define a new tool for financial analysis in this
study. 

Methodology-
 BIST Manufacturing Sector registered
190 companies for year 2015 and 173 companies for year 2016 are analyzed. Some
liquidity ratios, fiscal ratios, operational ratios and profitability ratios
are calculated and outlier companies are decided. Data Mining is the one of the
most important data processing tool. It can be used for clustering the data,
classification the data and defining variables that have similar behaviors. It
is tried to define a new financial analysis technique with combination of ratio
analysis and data mining. In this study, outlier detection and some clustering
algorithms are applied on BIST Manufacturing Sector registered companies.

Findings-
 BIST Manufacturing Sector registered
121 of 190 companies for year 2015 and 127 of 173 companies for year 2016 are
decided as outlier companies. These outlier companies might be evaluated for
sectorel researches or fraud detection etc. Companies are divided two clusters
with and without outlier companies for year 2015. In addition, companies are
divided four clusters with outlier companies and two clusters without outlier
companies for year 2016. Differences between the number of clusters and cluster
characteristics are related to economical conditions.

Conclusion-
In conclusion, Data Mining Techniques can be used as financial analysis
method, especially when we need to analyze many companies’ financial situation
at the same time. It is considered that sector characteristics, global and
local developments would indicate meaningful correlations with outlier
companies. Besides that, it is determined that universal thresholds values for
financial ratios (e.g. current ratio 2) are different for our country. These
values are calculated for our country and evaluated with sectorel, global and
local factors. 

Anahtar Kelimeler

Financial statement analysis, ratio analysis, data mining, clustering analysis, EM algorithm

Kaynakça

• Bames, Paul. Stock Market Efficiency. Insider Dealing and Market Abuse. Fanham. GB Gower, 2009, p.7
• Gürsoy, U. Tuğba Şimşek. Veri Madenciliği ve Bilgi Keşfi. 1. Baskı. Ankara. Pegem Akademi. 2009. s.1 Halka Arz ve Borsada İşlem Görme. Borsa İstanbul. 2016. http://www.borsaistanbul.com/data/kilavuzlar/Halka_arz_ve_borsada_islem_gorme.pdf (Erişim Tarihi 7 Ekim 2016)
• Krzysztof, Cios & Others. Data Mining: A Knowledge Discovery Approach. USA: Springer Science+Business Media. LLC. 2007, p.6
• Larose, Daniel T. Discovering in Data: An Introduction to Data Mining. USA. Wiley&Sons, Inc. 2005. p.1-2-3
• Myatt, Glenn J., and Johnson, Wayne P. Making Sense of Data I : A Practical Guide to Exploratory Data Analysis and Data Mining, USA. Wiley. 2014. p.19
• Peter,Chamoni. www.enzyklopaedie-der-wirtschaftsinformatik.de/...-/Data Eylül 2013 (Erişim Tarihi 7 Ekim 2016)
• Refaat, Mamdout. Data Preperation forData Mining Using SAS. San Francisco: Morgan Kaufmann Publishers. 2007. p.7-8
• Silahtaroğlu, Gökhan. Veri Madenciliği. 1. Basım. İstanbul.: Papatya Yayıncılık Eğitim. 2008. s.10
• Tuffery, Stephane. Data Mining and Statistics for Decision Making. United Kingdom. John Wiley & Sons Ltd. 2008. p.12
• Witten, Ian H. Frank, Eibe. Hall, Mark A. Data Mining: Practical Machine Learning Tools and Techniques. Morgan Kaufmann Series. USA. 2011. p.5