C.Jstoms, which is one of the three wings in Ethiopian Revenues and Customs Authorityrn(ERCA), is established to secure national revenues by controlling impons and exports as well asrncoll ecting go~emmental tax and duties. This research focuses on identification, modeling andrnanalysis of various conflicting issues that Ethiopian customs faces. One of the major problemsrnidentified during problem understanding is controlling and management of fraudu lent behaviorrnof fo reign traders. The declarams' intent to various types of fraudulent activities which result inrnthe need for serious inspection of declarations and al the same lime, the huge amount ofrndeclarations per day demand significant number of human resource and time.rnRecognizing this critical problem of the government, ERCA adopt Automated System forrnCustoms DAta (ASYCUDA). ASYCUDA attempts to minimize the problems through risk levelrnrecommendation to declarations using select ivi ty method that uses five parameters from therndecJarants' information. The fundamenta l problem to ASYCUDA risk leveling is, restricting thernvariables whi ch are used to assign risk level; this may lead to direct the declaration intornincorrect channel.rnThis research proposed a machine learning approach to model fraudulent behavior of importersrnthrough identification of appropriate parameters from the observed data to improve the qualityrnof service at Customs, ERCA. In this research, the researcher proposed automated fraudrndetection models which predict fraud behaviors of importing cargos, in which the problemrnassoc iated with ASYCUDA risk leveling wi ll be minimized. The models have been bui ltrnthrough machine learning techniques by using the past data which was collected from customsrndata of ERCA. The analysis has been done on inspected cargos records having 74,033 instancesrnand 24 attributes.rnFour different prediction models were proposed. The first model is fraud prediction model,rnwhich predicts whether incoming cargo is fraudulent or not. The second model is fraud categoryrnprediction model, which identifies the specific type of the fraud category among the tenrnidentified categories. The third model is fraud level prediction model. which class ifies the fraudrnlevel as high or low. The last model is fraud ri sk level prediction model which is used tornclassify the risk level of importing cargos into high. medium or low.rnxrni â€¢rnâ€¢rnMoreover. from the recommendation of IEEE, four best machine learning approaches have beenrntested for each of the identified prediction models. These are C4.5, CART. KNN and NaivernBayes. Based on the results which are obtained through various experimental analyses. C4.5 isrnfound to be the best algorithm to build all types of the prediction models. The accuracy obtainedrnin the first, second, third and founh scenarios using C4.5 machine learning algorithms arern93.4%,84.4%, 89.4%, and 86.8% respecti vely.rnThe next best algorithm, Classification and Regression Tree (CART), performed an accuracy ofrn92.9%,80. 1 %,89.4%,85.3% for the first, second, third and fourth scenarios respectively.rnThe researchers observed that both C4.5 and CART perform better for fraud prediction andrnfraud level classification compared to fraud category and risk level prediction. Moreover, NaivernBayes statistical approach is found to be very poor.rnKey words: Fraud prediction, fraud category prediction, fraud level prediction, fraud risk levelrnprediction, classification, machine learning algorithm, ASYCUDA.