Prediction of Financial Distress:

Analyzing the Industry Performance in Stock Exchange

Market using Data Mining
Harjani Rezkya Putri, Arian Dhini
Industrial Engineering Department
Faculty of Engineering, Universitas Indonesia
 Research Questions Results and
and Goals Conclusion
Outline

Research Research
Background Methods
Research
Background
Financial Distress Phenomenon
According to Platt &Platt (2002), financial distress is a declining firm’s financial condition,
happened before bankruptcy or liquidation.

Financial Distress Indicators Financial Fundamental

Business Expansion (Platt & Platt, 2002) Condition Analysis

• High operational cost Financial Ratios

• Lateness of salary
Financial Gaps • Decreasing sales volume
Liquidity Solvability
• Failing and growing liabilities
• Decreasing stock prices Ratio Ratio

Rentability Activity
Alternative Fund Ratio Ratio

How do firms get alternative funds to fulfill its financial obligations?

Indonesian Capital Market: Now
Increasing total number of capital market investor in Capital market includes several financial instruments,
Indonesia for the past years with stock investment being the most popular
Capital Market Investor [1]
Stock Transactions Average
2.000.000 1.617.367 450.000 422.496
386.696
1.500.000 1.122.668 400.000
894.116 324.736
350.000
1.000.000
434.107 300.000 264.127
500.000 320.506 364.465
250.000 212.635221.942
0 200.000 153.686
2013 2014 2015 2016 2017 2018
150.000 105.790113.454121.712
100.000
851.662 Stock
Investor 47% 53%
31,97% 50.000

From 2017 0
per Des’18 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019

Indonesia will be faced with the growing interest with stock investment

Does Indonesian firms are ready to face this growing trend of

stock investment?
[1] PT Bursa Efek Indonesia. (2019). PT Bursa Efek Indonesia. [online] Available at: https://www.idx.co.id/ [Accessed 11 Apr. 2019].
Investing activities brings many advantages in the form of shares, but that doesn’t
avoid investors from getting investing’s disadvantages
The dynamics and competing atmosphere in the stock
Dividen investment world caused some firms undergoes financial
Capital Gain distress condition,
2
forcing firms to do delisting
[1]
NUMBER OF DELISTED FIRMS
Capital Loss
9
Liquidation Risk 8

JUMLAH PERUSAHAAN
7 3
6
According to Wang (2014), the major 5
4 8 8
cause of delisting is a financial distress 3
7

condition in one firm 2

5
4
3
1
1
0
2009 2010 - 2012 2013 2014 2015 2016-
2011 2017
What could investors in Indonesia do to
assess firm’s financial condition in order to Delisting is a deletion of a firm in the stock market (go-
select the best finance-performing firms and private firms)
get the maximum amount of capital gain or
profit? [1] PT Bursa Efek Indonesia. (2019). PT Bursa Efek Indonesia. [online] Available at:
https://www.idx.co.id/ [Accessed 11 Apr. 2019].
Asikin, Z. (1997). Pokok-Pokok Hukum Perbankan di Indonesia. 6th ed. DKI Jakarta: Raja Grafindo Persada.
This research used several methods to build the FDP model

LogisticRegression
Logistic Regression DecisionTree
Decision TreeC4.5
C4.5 EnsembleClassifier
Ensemble Classifier

Analyze the relationship

between independent and Resulting a model based on Combines multiple classifiers
dependent variable with an a structure of a tree which to improve the classification
ordinal or nominal type of contains root, branch and performance of a single
data leafs. classifier.

- No restrictive - Could handle missing

assumption value - Higher performance
- Used to compare - Could handle continuous - Uses bagging and
conventional methods data boosting with each
with data mining - Visualization in the form advantages
methods of structured tree
Research Question
and
Purposes
This research aims to resolve the problems mentioned before by building the financial distress
prediction model

1 2
Only few financial distress prediction model that has been built does consider the use of
Research Gaps non financial variable and financial variable. Moreover, there is yet no conclusión of the
best model to predict the financial condition of firms in Indonesia with both non and
financial variable included

Key How is the FDP model performance with the use of methods chosen and the
Research participation of both non financial and financial variable?
Question

Built the most high permorming FDP model as an output to asses firm’s financial
Research condition with both non financial and financial variable which informations gathered
Purposes from Indonesia Stock Exchange (IDX) listed firms using data mining
Limitations of Research

Limitation Description

• Firms listed in Indonesia Stock Exchange with complete financial

report
• Financial variables used:
Limitations of ROE, ROA, EPS, PER, net profit margin, current ratio, debt to
condition, research equity ratio, quick ratio, working capital turnover rate,
period and object
operating profit margin.
Asumsi model cargo • Non financial variable used: age of firms
routing
• 5 years time period (2014-2018)
Research Methods
Data Collection

1. Return on Asset
2. Net Profit Margin
3. Return on Equity
4. Current Ratio
5. Earning per Share
Financial Variable 6. Debt to Equity Ratio
7. Price Earning Ratio Non Financial 1. Firm’s Age
8. Quick Ratio Variable
9. Working Capital Turnover
Rate
10. Operating Profit Margin
11. ART
12. Stock Price
Data Pre-Processing

Data Cleaning Checking and fixing missing values, inconsistent data.

Missing Value

Noisy Data

Data Scaling Transforming data into one desired range.

𝑥 − 𝑥min
𝑥=
𝑥𝑚𝑎𝑥− 𝑥min

Data in the year of 2014 until

Data Train 2017. Data Test Data in the year of 2018
Results and
Conclusion
 Classification Results

Prediksi Parameter LR DT C4.5 C4.5 + Prediksi

LR Non- Akurasi 87.32% 94.10% Non-
Distress
distress bagging Distress
distress
Precision 98.99% 91.28%
Distress 30 65 Recall 98.99% 96.60% Distress 64 31

Aktual
Aktual

Specificity 99.13% 91.91%

Non- Non-
8 473 distress
20 461
distress

Prediksi C4.5 + C4.5 + C4.5 + Prediksi

DT C4.5 Parameter
Distress
Non- bagging boosting boosting Distress
Non-
distress distress
Akurasi 100.00% 94.55%
Distress 64 31 Precision 100.00% 93.75% Distress 77 18

Aktual
Aktual

Recall 100.00% 94.66%

Non- Non-
distress
25 456 Specificity 100.00% 94.47% distress
13 468
 Model Performance Comparison
105,00%

100%
100,00% 98,99%
95% 95% 95%

95,00% 87,32% 86,40%

87,30%

90,00%

85,00% 93,75%
90,27% 91,14% 91,28% 90,30% 91,10% 90,10% 90,80%
80,00%

75,00%
LR DT C4.5 DT C4.5 DT C4.5 LR DT C4.5 DT C4.5 DT C4.5 LR DT C4.5 DT C4.5 DT C4.5 LR DT C4.5 DT C4.5 DT C4.5
+ + + + + + + +
boosting bagging boosting bagging boosting bagging boosting bagging
Accuracy Precision Recall Specificity

The graphic of model performance comparison shows that DT C4.5 completed with
boosting performed the best predictive model among LR, DT C4.5 and DT C4.5 with
bagging.
Model Performance Comparison

Kappa
Model AUC Score Description
Coefficient
AUC and Kappa Coefficient
LR 0.846 0.394 Excellent classification
shows DT C4.5 with ensemble
DT C4.5 0.835 0.637 Excellent classification classifier outperform other
DT C4.5 + Excellent classification models.
boosting 0.921 0.800
DT C4.5 + Excellent classification
bagging 0.940 0.662
Conclusion
Conclusion

This study applied conventional statistical methods compared to the data

1 mining techniques to predict the financially distressed firms, logistic
regression and decision tree along with the bagging and boosting.

The decision tree with boosting model showed the best prediction
performance with overall accuracy (94.61%), highest sensitivity
2
and specificity (94%) and has fewest errors in overall and type II
error rates (5.5%).

Return on Asset (ROA) being the most significant or

3
important predictive variable to determine financially distressed firms.
Future Research

The use of other data mining techniques to get the better predictive models

Expansion type of data collected, including a total of 9 sector in Indonesia Stock

Exchang listed firms.

Model constructed with the consideration of every sector’s characteristic

The use of other non-financial variables can be added and further investigate the
significant impact to predictive models
References
References
Alaka, H. A., Oyedele, L. O., Owolabi, H. A., Kumar, V., Ajayi, S. O., Akinade, O. O., & Bilal, M. (2018).
Systematic review of bankruptcy prediction models : Towards a framework for tool selection. Expert Systems
With Applications, 94, 164–184. https://doi.org/10.1016/j.eswa.2017.10.040
Altman, E. (1968). Financial Ratios, Discriminant Analysis and The Prediction of Corporate Bankruptcy, XXIII(No.
4), 21.
Asikin, Z. (1997). Pokok-Pokok Hukum Perbankan di Indonesia. 6th ed. DKI Jakarta: Raja Grafindo Persada.
Baesens, B., Vlasselaer, V., & Verbeke, W. (2015). Using Descriptive, Predictive and Social Network Techniques.
New Jersey: Wiley. Balcaen, S., & Ooghe, H. (2004). FACULTEIT ECONOMIE EN BEDRIJFSKUNDE
HOVENIERSBERG 24 B-9000 GENT 35 years of studies on business failure: an overview of the classical
statistical methodologies and their related problems, 32(0).
Chen, W. Sen, & Du, Y. K. (2009). Using neural networks and data mining techniques for the financial distress
prediction model. Expert Systems with Applications, 36(2 PART 2), 4075–4086.
https://doi.org/10.1016/j.eswa.2008.03.020
Dietrich, J. R. (1984). Discussion of Methodological Issues Related to the Estimation of Financial Distress
Prediction Models. Journal of Accounting Research, 22, 83. https://doi.org/10.2307/2490860
Dimitras, A. I., Zanakis, S. H., & Zopounidis, C. (1996). A survey of business failures with an emphasis on
prediction methods and industrial applications. European Journal of Operational Research, 90(3), 487–513.
https://doi.org/10.1016/03772217(95)00070-4
References
Doumpos, M., & Zopounidis, C. (1999). A Multicriteria Discrimination Method for the Prediction of Financial
Distress: The Case of Greece. Multinational Finance Journal, 3(2), 71–101. https://doi.org/10.17578/3-2-1
Geng, R., Bose, I., & Chen, X. (2015). Prediction of financial distress: An empirical study of listed Chinese
companies using data mining. European Journal of Operational Research, 241(1), 236–247.
https://doi.org/10.1016/j.ejor.2014.08.016
Gepp, A., & Kumar, K. (2015). Predicting Financial Distress: A Comparison of Survival Analysis and Decision Tree
Techniques. Procedia Computer Science, 54, 396–404. https://doi.org/10.1016/j.procs.2015.06.046
Harahap, S. S. (2009). Analisis Kritis atas Laporan Keuangan (14th ed.). RajaGraindo.
Hermawan, R., & Herusantoso, K. (2015). Penggunaan Teknik Data Mining untuk Memprediksi Financial Distress
pada Pemerintah Daerah Kabupaten/Kota di Indonesia, (2014), 3–4. Retrieved from
http://juliwi.com/published/E0205/PITIWI2-25.pdf
Khermkhan, J., Chancharat, N., Chancharat, S., & Theinthong, A. (2015). Differences in financial distress
prediction models for small and medium-sized enterprises. Kasetsart Journal - Social Sciences, 36(3), 533–
543.
Kim, S. Y., & Upneja, A. (2014). Predicting restaurant financial distress using decision tree and AdaBoosted
decision tree models. Economic Modelling, 36, 354–362. https://doi.org/10.1016/j.econmod.2013.10.005
References

Lee, T. S., Chiu, C. C., Chou, Y. C., & Lu, C. J. (2006). Mining the customer credit using classification and regression
tree and multivariate adaptive regression splines. Computational Statistics and Data Analysis, 50(4), 1113–
1130. https://doi.org/10.1016/j.csda.2004.11.006
Mantas, C. J., & Abellán, J. (2014). Credal-C4.5: Decision tree based on imprecise probabilities to classify noisy
data. Expert Systems with Applications, 41(10), 4625– 4637. doi:10.1016/j.eswa.2014.01.017
Rahmy, R. (2015). Pengaruh Profitabilitas, Financial Leverage, Sales Growth Dan Aktivitas Terhadap Financial
Distress (Studi Empiris pada Perusahaan Manufaktur yang Terdaftar di BEI Tahun 2009-2012.
Romadhoni. (2017). Pengaruh Struktur Modal Terhadap Kinerja Keuangan Perusahaan Manufaktur Sektor
Makanan Dan Minuman Yang Terdaftar Di Bursa Efek Indonesia Tahun 2013-2016. Jurnal Riset Manajemen,
000(November 2016), 219– 232.
Salehi, M., Shiri, M. M., & Pasikhani, M. B. (2014). Predicting corporate financial distress using data mining
techniques: An application in Tehran Stock Exchange. International Journal of Law and Management, 58(2),
216–230. https://doi.org/10.1108/IJLMA-06-2015-0028
Santosa.Budi, U. A. (2018). Data Mining dan Big Data Analitycs.
Tsai, C. F. (2014). Combining cluster analysis with classifier ensembles to predict financial distress. Information
Fusion, 16(1), 46–58.
 Thank You
[email protected]
[email protected]
Decision Trees vs SVM

Hastie et al.,”The Elements of Statistical Learning: Data Mining, Inference, and Prediction”, Springer (2009)
Bagging Example
Bagging

x2
• Reduces overfitting (variance)
• Normally uses one type of classifier
• Decision trees are popular
• Easy to parallelize

x1