TY  - SER
AU  - Ei Thaw Win
AU  - Guha,Sumanta
AU  - Vatcharaporn Esichaikul,
AU  - Duboz,Raphael
ED  - Ministry of Foreign Affair, Norway,
TI  - Applying data mining techniques: case studies to predict loan defaults and vehicle quality
T2  - Research studies project report
PY  - 2015///
CY  - Pathum Thani, Thailand
PB  - Asian Institute of Technology
KW  - Data mining
KW  - Used cars
KW  - Data processing
N1  - A research submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Computer Science, School of Engineering and Technology; Research Studies Project Report (M.Eng.) - Asian Institute of Technology, 2015
N2  - Data mining is popularly applied to discover ground truths from huge amounts of data.Data mining  techniques  in  this  study  are  applied  to  two  different  problems  to  study  the effectiveness  of  DM  tools  with  different  characteristics  from  the  Bank  Industry  and  Used Cars Market, respectively.The  first  problem  to  tackle  is the  recovery  of  loans  in  the  banking  industry  that  will predict the expected percentage of return on future loans.The  second  problem  considers  the  automobile  industry.  It  deals  with  car  dealerships purchasing used vehicles at an auction and predicts the likelihood that the vehicle will be a bad buy.The objectives of this research study is applying data mining tools and techniques to solve the  above  problems  and  determine  the  best  method  suitable  for  the  characteristics  of  the problem.In this research paper, data mining models which include k-NN, random forest and support vector machine (SVM) are applied on two different datasets.The goal of this study was to compute the performance of data mining techniques on both problems: binary and multiple classification using the evaluation techniques: classification accuracy, precision, recall and root mean square error.According to the experimental results, the SVM model verifies to have the best performance in  both  problems;  loan  default  prediction  and  vehicle  quality  prediction.  Random  forest model also performs well in both datasets. The performance of K-NN algorithm is better in vehicle quality prediction than loan default prediction
UR  - http://203.159.5.9/ait-thesis/detail.php?q=B01315
ER  -