et

International Journal on Emerging Technologies 15(2): 05-12(2024)

ISSN No. (Print): 0975-8364
ISSN No. (Online): 2249-3255

Model for the Prediction of Default Risk of Funding Requests Using Data Mining
Sameh Ali1*, Atef Raslan2 and Lamiaa Fattouh3
1
Faculty of Graduate Studies for Statistical Research Cairo University Cairo, Egypt.
2
Department of Information Systems, Higher Institute of Advanced Studies.
3
Faculty of Graduate Studies for Statistical Research Cairo University Cairo, Egypt.
(Corresponding author: Sameh Ali*)
(Received 02 May 2024, Revised 10 June 2024, Accepted 15 July 2024)
(Published by Research Trend, Website: www.researchtrend.net)

ABSTRACT: Microfinance institutions currently confront numerous financing challenges, particularly

within the non-bank sector where risks abound. Each year, there is a notable incidence of borrowers
defaulting on their microfinance obligations, resulting in substantial financial setbacks for these companies.
Given the burgeoning volume of electronic data and transactional activity in the banking realm, data mining
emerges as a pivotal strategic domain. Leveraging data mining techniques, valuable patterns and insights can
be gleaned from vast datasets, thereby furnishing actionable information to mitigate risks associated with
nonbank loans. This study employs data mining as a tool to extract pertinent insights from the credit data of
microfinance companies, facilitating the construction of a model aimed at assessing borrower eligibility and
identifying potential default risks. The study employs the open-source machine learning platform WEKA.
This study uses data mining to develop a predictive model for microfinance institutions to enhance decision-
making in client financing. By employing cross-validation and percentage splits (80-20, 70-30, 60-40), cross-
validation showed slightly higher accuracy. The model performed excellently, especially with preprocessed
data, highlighting the importance of data cleaning. The J48 proved to be the most effective algorithm,
demonstrating superior accuracy. The study emphasizes the potential of using historical data to assess client
credit status during financing approvals, reducing loan defaults, and supporting the growth of non-banking
institutions.
Keywords: Data Mining Technique, Classification, Credit Risk, Non-Banking Sector, Microfinance, Fraud
detection.

INTRODUCTION credit history, employment status, and demographic

profile, data mining can discern the credit behavior of
Microfinance institutions and non-governmental
borrowers. This information enables microfinance
organizations are established to provide microcredit to
companies to assess customers and determine their
economically disadvantaged individuals who lack
suitability for a loan, as well as identify potential
stable employment or credit history, making it difficult
default risks. Armed with insights into a borrower's
for them to access traditional banking services. These
likelihood of default, microfinance companies can
borrowers often engage in small-scale economic
proactively mitigate risks, thereby enhancing their risk
activities but are underserved by conventional banks.
management practices (Subia and Galapon 2020; Zeng
Egypt hosts numerous recognized microfinance
et al., 2017).
institutions, with "Tamweely Microfinance" being one
Utilizing data mining techniques, it becomes possible
of the accredited institutions in the country
to analyze the behavior and reliability of borrowers
(Hailemariam et al., 2012). The rapid progression of
associated with microfinance companies. However,
science and information technology has led to the
ensuring the quality of data is paramount, as the
generation and continual maintenance of vast volumes
efficacy of decisions hinges upon it. Through
of data. Data mining, utilizing fundamental techniques
meticulous data preprocessing, the raw experimental
for extracting information and identifying patterns, has
data can be refined, allowingfor the selection of
emerged as a crucial tool across various fields of
variables that contain pertinent and essential
endeavor. Notably, within the non-banking sector, data
information exclusively. Real-world data often contains
mining applications play a significant role. One such
noise, inconsistencies, redundancies, or irrelevant
application is credit scoring, among the earliest tools
information, necessitating the elimination of outliers,
developed for financial risk management.
standardization, and data cleaning to attain the desired
Credit scoring provides valuable insights to lenders in
data quality (Abakar, 2020).
the banking industry, aiding in lending decisions.
One of the most significant challenges in data mining
Furthermore, data mining empowers microfinance
research pertains to the initial stage of refining gathered
companies to enhance decision-making throughout the
data, particularly in large databases. Achieving accurate
loan approval process. By analyzing factors such as
and useful results mandates that the data be both
Ali et al., International Journal on Emerging Technologies 15(2): 05-12(2024) 5
relevant and original. Handling such data is both time- enabling business analysts to unearth valuable patterns
consuming and crucial. Moreover, clear correlations within available data. Additionally, data preprocessing
must exist within the data to yield valuable patterns. plays a crucial role in data mining, as the quality and
Enhancing the efficiency and accuracy of classifications completeness of data significantly impact the
further requires the incorporation of algorithms to effectiveness of data mining algorithms (Zeng et al.,
develop a meaningful model (Kulkarni and Kulkarni 2017).
2016). Sophisticated data mining technologies, tailored to
In light of these considerations, this paper outlines a various data sets and objectives, facilitate the prediction
systematic approach to creating a sample model for of outcomes for novel scenarios based on patterns
predicting default risk. identified from familiar instances. Such predictive
This research delineated the following objectives: capabilities offer insights into potential future outcomes
Investigate the viability of employing data mining of implementing a strategy and enable risk assessment.
techniques to unearth patterns within typical Big Data has emerged as a valuable resource for
microfinance datasets. Construct a predictive model microfinance firms, serving as a tool for assessing
leveraging these discerned patterns to identify potential creditworthiness and detecting fraudulent activities
microfinance defaulters. Identify key financial (Nguyen, 2019).
parameters and successful project attributes conducive
REVIEW OF RELATED LITERATURE
to averting defaults. Facilitate streamlined decision-
making by early detection of project success or failure. This section outlines the literature review and pertinent
Present findings and offer recommendations for future research endeavors. Various studies have explored
research endeavors. similar themes, and below is a concise overview of
The scope of this research is confined to exploring the several papers we have examined and scrutinized.
potential of data mining techniques in forecasting The author introduced a novel methodology for
defaulters to enhance company performance and evaluating loan risk within the banking sector,
facilitate informed decision-making. The focus leveraging data mining techniques. To predict loan
primarily revolves around classification methods, statuses, the model was constructed using data sourced
deemed suitable for constructing a predictive model from the banking industry. The dataset comprises 1000
aimed at extracting insights from non-banking instances, partitioned into a training set (80% of the
investment data. The dataset under scrutiny will be data) and a testing set (20% of the data). Three
sourced from Tamweely Microfinance Company in algorithms—Bayes Net, J48, and Naïve Bayes—were
Egypt. employed to develop the proposed model, implemented,
and evaluated using the WEKA software. The accuracy
BACKGROUND
measures for each algorithm were as follows: Bayes
Effectively managed and modeled data possesses the Net (73.8739%), J48 (78.3784%), and Naïve Bayes
capacity to provide valuable insights that greatly (73.8739%). The results were extensively discussed,
enhance decision-making processes. Within the and a comprehensive comparison of the algorithms was
financial industry, data warehouses serve as the conducted. Based on the findings, J48 was selected as
bedrock, providing organizations with the means to the most accurate algorithm (Abakar, 2020).
harness data for informed decision-making. Through The primary objective of the study was to explore the
meticulous analysis of this data, organizations can make application of data mining techniques in examining
well-founded evaluations concerning the feasibility and customer loyalty and forecasting loan default
potential success of projects seeking financial backing. occurrences. This encompassed identifying strategies
Furthermore, delving into historical data allows for the for customer retention, forecasting liquidity risk,
identification of projects that have demonstrated introducing novel services, and enhancing profit
success in the past, thereby assisting in risk mitigation margins, rated on a scale from 1 to 5. Through
for both the funding institution and the applicants (Zeng experiments conducted on a dataset comprising 9551
et al., 2017). records extracted from a database, it was found that the
It signifies that leveraging data-driven approaches could J48 classifier algorithm yielded favorable outcomes in
present fresh opportunities for enhancing business accurately classifying instances. The study utilized the
strategies. Data mining, also referred to as Knowledge WEKA data mining tool for analysis (Hailemariam et
Discovery, involves extracting significant, non-obvious, al., 2012).
implicit, previously unidentified, and potentially The case study introduces the application of various
valuable information or patterns from extensive data mining technologies in crafting a mechanism for
databases. The significance of data mining has been evaluating loan risk tailored for a subprime lender.
evident for over twenty years. Forward-thinking Diverse data mining methods were employed to derive
enterprises are prioritizing data in their strategic the outcomes, with the analysis conducted using the
decision-making processes. Data analysis typically WEKA data mining tool. The dataset comprises 1000
progresses in two stages: Discovery and search. The instances, consisting of 700 good cases and 300 bad
patterns discovered in the discovery phase can be cases. The experimental process involved training the
utilized in the subsequent search phase. It's important to models on 70% of the dataset and testing them with the
note that data mining isn't mere data reporting. Genuine remaining 30%. The algorithms employed include J48,
data mining serves specific purposes and is a statistical EM, Naïve Bayes, K-means. The accuracy rates for
process aimed at achieving business objectives, each method were determined as follows: J48 achieved
Ali et al., International Journal on Emerging Technologies 15(2): 05-12(2024) 6
71.44%, EM yielded 28.77%, Naïve Bayes attained To ensure precision, significant factors determining the
75.09%, and K-means reached 57.47%. It was creditworthiness of applicants were incorporated into
concluded that a decision tree, represented by the J48 the model, including installment type, age, monthly
algorithm, is the most suitable data mining technology expenses, job sector, payment method, and income-to-
for developing a loan risk assessment system tailored finance ratio. By presenting a systematic.and structured
for subprime lenders (Lee and Wang 2020). approach to developing a credit scoring model, this
This study constructed a model aimed at evaluating and study contributes to advancing credit scoring
determining the suitability of a borrower for a loan or methodologies. Based on the study's findings, banks
assessing the risk of default. The model, implemented can utilize this model to construct their credit scoring
using the WEKA software, underscores the crucial role systems for assessing the creditworthiness of individual
of preprocessing or data cleaning in enhancing accuracy loan applicants. Implementing this model can help
rates. Particularly, the results achieved through the J48 banks mitigate risks and enhance long-term operational
algorithm were noteworthy, showcasing a high efficiency in the credit system, facilitating informed
correctly classified instances rate of 96.3647%. The decision-making processes.
dataset utilized comprised 3466 instances with The model empowers loan officers to automate
attributes structured in the format (AIRFF) (Nalić and decision-making and accurately predict the
Švraka 2018). creditworthiness of applicants. However, the study
The study utilizes data mining techniques to enhance acknowledges the exclusion of several variables during
bank performance and decision-making by predicting the model's development, including net worth,
defaulters. Experiments were conducted using education level, number of dependents, other financial
microfinance data obtained from an agricultural bank in commitments, financing duration, and gender.
Sudan to forecast microfinance status. The Random Incorporating these variables in future iterations could
Forest, KNN, and Naive Bayes classification algorithms potentially enhance the model's accuracy and predictive
were employed, yielding different accuracy rates: capabilities. Additionally, expanding the dataset to
Random Forest achieved 94.6%, KNN reached 87.4%, include information from other banks can further
and Naive Bayes attained 92.3%. Based on these enhance the model's accuracy.
accuracy rates, Random Forest was selected as the Lastly, the study emphasizes the importance of ongoing
optimal algorithm. The Orange application data mining model maintenance to ensure its continued reliability in
tool was utilized for analysis, with recommendations the dynamic business environment. Regular updates
suggesting the monitoring of funded projects from and adjustments are crucial to adapt the model to
inception to mitigate default risks (Han et al., 2012). evolving market conditions and changing customer
The study aims to explore factor analysis, data mining profiles (Sum et al., 2022).
modeling, credit scoring, and post-modeling processes.
Table 1: Summarize the literature review.
Authors
Title of paper Methodology Result
“year”
Bayes Net, J48, and Naïve
Bayes Net 73.87%, J48
Randula Koralage, Data Mining Techniques for Bayes the model was
78.38%, and Naïve Bayes
“2019” Credit Card Fraud Detection implemented and evaluated
73.87%.
using the WEKA software
Using data mining approaches to
J48 classifier algorithm
Jasmina Nalić and build credit scoring model: Case
J48 classifier algorithm using performed relatively well in
Amar Švraka, study-implementation of credit
the WEKA software accurately classifying
“2018” scoring model in microfinance
instances
institution.
GLM algorithm, demonstrated
Jia Wu, Sunil Vadera,
excellent results, with a
Karl Dayson, Diane
A comparison of data mining GLM algorithm using the Oracle predictive confidence of
Burridge and Ian
methods in microfinance Data Mining (ODM) software 97.437% and an overall
Clough
average accuracy exceeding
“2010”
98%
Jun-Ya Zeng, Jian- A new competing risks model
J48 71.44%, EM 28.77%,
Bang Lin and Tian for predicting prepayment and J48, EM, Naïve Bayes, K-means,
Naïve Bayes 75.09%, and K-
Wang default using data mining Using WEKA software
means 57.47%.
“2017”
Dr. Md. Rashid Displaying a correctly
Effectiveness of Data mining in
Farooqi and Naiyar J48 algorithm Using WEKA classified instances rate of
Banking Industry: An empirical
Iqbal software 96.36% on a dataset with
study
“2017” 3,466 instances
Rabihah Md, Waidah
A New Efficient Credit Scoring
Ismail, Zul Hilmi Random Forest, KNN, and Naive
Model For Personal Loan Using Random Forest 94.6%, KNN
Abdullah, and Nurul Bayes classification algorithms
Data Mining Technique For 87.4%, and Naive Bayes
Fathihin Mohd Noor Using The Orange data mining
Sustainability Management 92.3%.
Shah tool
“2022”
Ali et al., International Journal on Emerging Technologies 15(2): 05-12(2024) 7
After reviewing previous research and studies, we've A detailed plan is devised to accomplish both the data
identified an intriguing area for further investigation: mining and business goals, encompassing the
exploring how non-bank financial institutions can preliminary selection of tools and methodologies
effectively integrate traditional and digital methods (Figueiredo et al., 2023). from a business angle.
seamlessly. While existing studies primarily focused on Nevertheless, there are situations where subjective
challenges within the banking sector, such as payment criteria are vital, like delivering valuable insights into
defaults, installment failures, and risk assessment for customer relationships. Data Mining Success Criteria:
decision-making, there's a noticeable gap regarding This outlines the milestones for a successful outcome
similar issues within non-banking sectors. from a data mining viewpoint. For instance, attaining a
During our search, we found a lack of research or specific level of predictive accuracy in the model.
scientific papers addressing these points within non-
DATA UNDERSTANDING
banking sectors, including companies and non-
governmental organizations catering to customers who Data understanding and preparation are crucial factors
may lack the financial capabilities or face procedural influencing the results of data mining efforts. The
complexities typical in traditional banking settings. In effectiveness of the constructed model significantly
contrast, these entities often offer faster financing depends on the depth and accuracy of data acquisition,
solutions, sometimes disbursing funds within 24 hours. examination, and preprocessing (Jackson, 2002). Thus,
Given Egypt's current focus on financial inclusiveness the following sections explore data understanding and
in the digital age and the proliferation of microfinance the essential preprocessing tasks performed in this
companies and non-governmental organizations in the context.
country, there's an opportunity to explore how data Since the data typically originates from routine
mining algorithms can address specific challenges faced transactions collected for administrative purposes, it's
by these entities. One such challenge is accurately crucial to evaluate the existing data landscape to
estimating the success of projects submitted for identify relevant aspects and understand its nature.
financing and predicting their likelihood of success or In this context, the dataset was obtained from
failure to mitigate non-payment or installment defaults. Tamweely Microfinance Institution, categorized as
social data, and gathered from various branches
DATA MINING PROCESS
spanning from 2018 to 2023. Initially, the primary
The aim is to construct a practical predictive model objective was to consolidate the data into a unified
through data mining techniques and represent the repository, resulting in the accumulation of 534,639
outcomes in a visually comprehensible manner. Fig. 1 records for preprocessing tasks. Although the dataset
illustrates the sequential steps of the data mining comprised 19 attributes in total, certain attributes
process utilized in this particular research investigation. contained numerous missing values, noises, and
inconsistencies requiring resolution during the data
cleaning phase of data preparation.
DATA PREPARATION
During this phase, data preparation involves various
processes applied to the extracted data to enhance its
suitability for the experiment and improve the overall
data mining task. At this stage, the most crucial
preprocessing tasks were carried out. These include
data selection, data cleaning, and data
aggregation/summarization, as outlined below.

Fig. 1. The Cross-industry standard process for data

mining.

BUSINESS UNDERSTANDING
In the initial phase of the CRISP-DM process, the
primary focus is on comprehending the business
objectives and constraints, maintaining equilibrium
amidst various priorities within the organization,
Tamweely. This stage is dedicated to pinpointing the Fig. 2. Data Preparation.
Business Success Criteria: Here, clear and quantifiable
standards are set to evaluate the success of the model
pivotal factors that impact the data analysis objectives.
Ali et al., International Journal on Emerging Technologies 15(2): 05-12(2024) 8
DATA SELECTION Table 3: Information about the data set (Conditional
Attribute).
The main aim of data selection is to identify the suitable
data type, source, and instrument(s). Initially, data Attribute Description
reduction involves eliminating unnecessary or less BRCODE Branch Code
DEBT_TYPE Product Type
important attributes from the original dataset. This
PRIMIUM_VALUE The value of Installment
process is based on the objective of the study at hand. DEBT_PRD Funding duration in months
Microfinance data is obtained from Tamweely APPROV_VALUE Funding value
Microfinance from 2018 to 2023. In this dataset, there RATE Annual interest
are 534639 instances, and 19 Attribute the Table 1 APP_FEE Application submission fees
gives information about the data set. TOTAL_REQ_
Total Funding
AMTOUNT
DATA CLEANING The payment period is in
REQ_NO_MONTHS
months
After attribute selection, the next step is data cleaning, TOTAL_AMOUNT Total Funding with interest
which is applied to the dataset with the selected INDUSTRY_CODE Industry code
MAIN_ACTIVITY Main activity code
attributes. Data collected for the mining process often ACTIVITY_TYPE Sub-activity code
contains missing values, noise, or inconsistencies, GOV_ID Governorate code
which can result in the generation of unreliable JOBCODE Job Code
information during the mining process. A high-quality EDUCATION_CODE Education Code
data mining process typically produces efficient results, SEX Gender
requiring preprocessing of the collected data to enhance SCORE Credit inquiry
OPEN_CREDITS Number of open Funding
its quality and, consequently, the mining outcomes
(Aljawarneh et al., 2019).
In this study, various standard data preprocessing tasks
are conducted on the dataset, including data integration,
data cleaning, data reduction, and data transformation.
The initial step of data preprocessing involves Data
Filtering, where relevant attributes necessary for
prediction are selected from the company dataset. Since
the dataset is unorganized, with features nested within
each other, efforts a Are made to rearrange similar
fields together to ensure accuracy. For example, all
features related to monetary details are grouped, and
likewise for premium-related features.
The subsequent task is handling missing data. The
dataset contains missing and imputed data.which are
addressed in this step. For instance, missing data in
attributes such as "Total Amount," "Main Activity
code," "Activity Type Code," and "Education Code" are Fig. 3. Missing value [Total Amount].
handled by replacing the missing value with the mean
of all samples belonging to the same class as the given
tuples. As shown in the following figure (3,4,5,6).
Filling in missing values, and removing inconsistencies
and noises were major data-cleaning activities
undertaken at this stage of data preparation. Some fields
had missing values, with 12,000 missed values from the
"Total Amount" field and 7,600 missed values from the
"Education Code" field. These values were considered
most probable because they had the highest mode in the
original dataset.
Table 2: Information about the data set (Class
Attribute).
Flag Description
Good Disbursement of funding
Bad Customer Reject
V-Bad Reject final

Fig. 4. Replacing missing value [Total Amount].

Ali et al., International Journal on Emerging Technologies 15(2): 05-12(2024) 9

 implementation and evaluation of model-building
experiments. We opted for the J48 classification
algorithm to build the model to achieve higher
precision. A tree classifier is particularly effective in
determining whether an individual is a suitable
applicant for a loan or if there's a high risk of default.
TYPES OF EXPERIMENT
A. Cross-Validation Method
During this experiment, the J48 Tree algorithm was
implemented on the dataset utilizing all features and
instances. The experiment employed 10-fold Cross-
Validation, as depicted in Figure 7. The achieved
accuracy was 99.7901%.

Fig. 5. Missing value value [Education code].

Fig. 7. J48 Tree (10-fold Cross-Validation.

B. Percentage Split Method
In this experiment, the J48 Tree algorithm was utilized
on the dataset incorporating all features and instances.
The experiment was repeated several times, altering the
sizes of the training and test sets (80% training, 20%
test - 70% training, 30% test - and 60% training, 40%
Fig. 6. Replacing missing value [Education code]. test). The most favorable outcome was observed when
MODELLING the data was partitioned into 60% training and 40% test
sets. The achieved accuracy was 99.79%.
This study employed the J48 algorithm, a classification
model utilized in data extraction within the Weka Table 4: Information about the Accuracy for J48
toolkit. Tree.
Training Test Accuracy
CLASSIFICATION J48Tree 80% 20% 99.7681%
Two methodologies for data analysis are employed to algorithm 70% 30% 99.7793%
60% 40% 99.79%
create models for identifying significant categories and
predicting future data patterns. These methodologies
are referred to as Classification and
Prediction.Classification models are crafted to predict
categorical class labels, while prediction models are
devised to predict continuous valued functions. For
instance, a classification Model could be constructed to
categorize bank loan applications as either safe or
risky. Prediction involves the model's capability to
accurately forecast the classification of incoming data.
It assesses whether the model can appropriately classify
the new data (Han et al., 2012).
EXPERIMENTS RUN
The experiment has been carried out. This section
delineates the various activities conducted regarding the Fig. 8. J48 Tree (80% training, 20% test).
Ali et al., International Journal on Emerging Technologies 15(2): 05-12(2024) 10
 Fig. 12. Re-evaluation test (Bod).

Fig. 13. Re-evaluation test (V-Bod).

EVALUATION AND DATA PRESENTATION
Fig. 9. J48 Tree (70% training, 30% test). In this study, two approaches were employed to train
and assess the model (cross-validation and percentage
splits). As mentioned earlier, the main objective of
developing the classification model is to discern
patterns in each borrower's status, aiding in the
prediction of a new borrower's status based on these
characteristics. Various classification trees were
examined using the J48 classifier algorithm, with the
model achieving the highest accuracy among them.
This experiment yielded the most favorable outcome
compared to all other experiments conducted, primarily
due to its superior accuracy level. Two methodologies
were employed for training and testing the model:
cross-validation and percentage splits (80-20, 70-30,
60-40). The findings indicated that cross-validation
resulted in slightly higher accuracy compared to
percentage splits, with an accuracy rate of 99.7901%.
Overall, the accuracy at the model level was highly
satisfactory. Precision peaked, and the confusion matrix
displayed commendable outcomes. Furthermore, the
classifier demonstrated robust performance when
applied to a preprocessed dataset.
CONCLUSIONS
The paper utilizes data mining to develop a predictive
Fig. 10. J48 Tree (60% training, 40% test). model, focusing on the loan histories of existing
C. The result test of the experiment borrowers. This model aims to aid in comparing
potential loan applications by identifying characteristics
indicative of a good or bad loan record, drawing from
credit background and demographic profiles. Emphasis
is placed on the importance of preprocessing or
cleaning data to achieve higher accuracy rates. The
results obtained using the J48 algorithm are particularly
noteworthy, with a correctly classified instances rate of
99.7901%. Additionally, the preprocessing stage can
reveal patterns useful for identifying target loan
markets, devising income-enhancing strategies,
Fig. 11. Re-evaluation test (Good). reducing default risk, and improving loan products.

Ali et al., International Journal on Emerging Technologies 15(2): 05-12(2024) 11

FUTURE SCOPE Management of Emergent Digital EcoSystems (pp.
183-184).
Future research could further investigate advanced Han, J., Kamber, M., & Pei, J. (2012). Data mining concepts
techniques in data mining and predictive analytics, and techniques third edition. University of Illinois at
aiming to refine the methodologies used and enhance Urbana-Champaign Micheline Kamber Jian Pei Simon
their applicability across different sectors. Exploring Fraser University.
these avenues will provide a more comprehensive Jackson, J. (2002). Data Mining; A Conceptual Overview.
understanding of customer behavior and microfinance Communications of the Association for Information
Systems. Communications of the Association for
dynamics, ultimately contributing to more effective and
Information System, 8.
strategic business practices. Kulkarni, E. G., & Kulkarni, R. B. (2016). Weka powerful
Acknowledgements. First of all, I would like to extend tool in data mining. International Journal of Computer
my special gratitude to Dr. Atef Raslan for his Applications, 975, 8887.
Lee, Y. C., & Wang, Y. (2020). Credit risk analysis of
unreserved support and assistance throughout my
microcredit using the data of a chinese microfinance
research work. His constructive comments and company and data mining techniques. ICFE 2020,
suggestions, in my study and in writing the thesis, were 394.
highly valuable. In general, his helpful personality, Nalić, J., & Švraka, A. (2018). Using data mining approaches
cooperativeness, and dedication are remarkably worth to build credit scoring model: Case study—
mentioning. implementation of credit scoring model in
My special thanks also go to my supervisor Prof. Dr. microfinance institution. In 2018 17th International
Lamiaa Fatouh for his continued efforts during all Symposium Infoteh-Jahorina (INFOTEH) (pp. 1-5).
phases of the research. IEEE.
Nguyen, C. (2019). The credit risk evaluation models: An
REFERENCES application of data mining techniques, Association for
Information Systems AIS Electronic Library (AISeL).
Abakar, M. I. (2020). Microfinance Data Analysis in Banking Subia, M. G., & Galapon, A. C. (2020). Sample model for the
Sector Using Data Mining Techniques) Case Study: prediction of default risk of loan applications using
Agricultural Bank of Sudan). Sudan University. data mining. Journal of Journal of Scientific &
Aljawarneh, S., Yassein, M. B., & Aljundi, M. (2019). An Technology Research, 9(6), 222-227.
enhanced J48 classification algorithm for the anomaly Sum, R. M., Ismail, W., Abdullah, Z. H., Fathihin, N., Shah,
intrusion detection systems. Cluster Computing, 22, M. N., & Hendradi, R. (2022). A New Efficient Credit
10549-10565. Scoring Model for Personal Loan using Data Mining
Figueiredo, R., Magalhães, C., & Huber, C. (2023). How to Technique Toward for Sustainability Management.
predict the innovation to SMEs? Applying the data Journal of Sustainability Science and Management,
mining process to the spinner innovation 17(5), 60-76.
model. Social Sciences, 12(2), 75. Zeng, J. Y., Lin, J. B., & Wang, T. (2017). A new competing
Hailemariam, G., Hill, S., & Demissie, S. (2012). Exploring risks model for predicting prepayment and default
data mining techniques and algorithms for predicting using data mining. In 2017 2nd International
customer loyalty and loan default risk scenarios at Conference on Image, Vision and Computing (ICIVC)
wisdom microfinance, Addis Ababa, Ethiopia. In (pp. 985-989). IEEE.
Proceedings of the International Conference on

How to cite this article: Sameh Ali, Atef Raslan and Lamiaa Fattouh (2024). Model for the Prediction of Default Risk of Funding
Requests Using Data Mining. International Journal on Emerging Technologies, 15(2): 05–12.

Ali et al., International Journal on Emerging Technologies 15(2): 05-12(2024) 12