At present, financial fraud detection is interested by many machine learning researchers. This is because of existing a big ratio between normal transactions and abnormal ones in data set. Therefore, a good result of prediction rate does not mean that there is a good detection result. This is explained that the experimental result might be effected by the imbalance in the dataset. Resampling a dataset before putting to classification process can be seen as the required task for researching in financial fraud detection area. An algorithm, so-called as MASI, is proposed in this paper in order to improve the classification results. This algorithm breaks the imbalance in the data set by re-labelling the major class samples (normal transactions) to the minor class ones basing the nearest neighbor’s samples. This algorithm has been validated with UCI machine learning repository data domain. Then, the algorithm is also used with data domain, which is taken from a Vietnamese financial company. The results show the better in sensitivity, specificity, and G-mean values compared to other publication control methods (Random Over-sampling, Random Under-sampling, SMOTE and Borderline SMOTE). The MASI also remains the training dataset whereas other methods do not. Moreover, the classifiers using MASI resampling training dataset have detected better number of abnormal transactions compared to the one using no resampling algorithm (normal training data).

Trends of currency rates can be predicted with supporting from supervised machine learning in the transaction systems such as support vector machine (SVM). By assumption of binary classification problems, the SVM can predict foreign exchange transaction as uptrend or downtrend. The prediction is performed basing on collected historical data. Alternative SVM models have been used to vote the best one, which is deployed detail in Expert Advisor (Robotics). This is to show that support vector machine models might help investors to automatically make transaction decisions of Bid/Ask in Foreign Exchange Market. For comparison, the transactions without using SVM model also are performed. The results of experimental transactions show the advantages of using SVM model compared to the transactions without using SVM model.

In recent years, the mining research over data stream has been prominent as they can be applied in many alternative areas in the real worlds. In this paper, we have proposed an algorithm called MFIWDSIM for mining frequent itemsets with weights over a data stream using Inverted Matrix [10]. The main idea is moving data stream to an inverted matrix saved in the computer disks so that the algorithms can mine on it many times with different support thresholds as well as alternative minimum weights. Moreover, this inverted matrix can be accessed to mine in different times for user's requirements without recalculation. By analyzing and evaluating, the MFIWDSIM can be seen as the better algorithm compared to WSWFP-stream [9] for mining frequent itemsets with weights over data stream.

In recent years, the mining research over data stream has been prominent as they can be applied in many alternative areas in the real worlds. In [20], a framework for mining frequent itemsets over a data stream is proposed by the use of weighted slide window model. Two algorithms of single pass (WSW) and the WSW-Imp (improving one) using weighted sliding model were proposed in there to solve the data stream problems. The disadvantage of these algorithms is that they have to seek all data stream many times and generate a large set of candidates. In this paper, we have proposed a process of mining frequent itemsets with weights over a data stream. Based on the downward closure property and FP-Growth method [8,9] an alternative algorithm called WSWFP-stream has been proposed. This algorithm is proved working more efficiently regarding to computing time and memory aspects.