A two-phase feature selection technique using mutual information and XGB-RFE for credit card fraud detection

Priscilla, C. Victoria; Prabha, D. Padma

doi:10.19101/ijatee.2021.874615

Cited by 14 publications

(3 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The outcome of the proposed model exhibits a promising f1-score of 91%. The experimental evaluation results of AE-XGB were compared with other related methods such as AdaBoost+XGB [48], Deep-Q NR [49] and XGBoost [50] of same dataset. The comparison result shows that AE-XGB with 𝜃=0.3 had the good precision score of 91% and high recall of 90% with a highest f1-score of 91%.…”

Section: Discussionmentioning

confidence: 99%

Probabilistic XGBoost Threshold Classification with Autoencoder for Credit Card Fraud Detection

Prabha,

Priscilla

2023

IJRITCC

View full text Add to dashboard Cite

Due to the imbalanced data of outnumbered legitimate transactions than the fraudulent transaction, the detection of fraud is a challenging task to find an effective solution. In this study, autoencoder with probabilistic threshold shifting of XGBoost (AE-XGB) for credit card fraud detection is designed. Initially, AE-XGB employs autoencoder the prevalent dimensionality reduction technique to extract data features from latent space representation. Then the reconstructed lower dimensional features utilize eXtreame Gradient Boost (XGBoost), an ensemble boosting algorithm with probabilistic threshold to classify the data as fraudulent or legitimate. In addition to AE-XGB, other existing ensemble algorithms such as Adaptive Boosting (AdaBoost), Gradient Boosting Machine (GBM), Random Forest, Categorical Boosting (CatBoost), LightGBM and XGBoost are compared with optimal and default threshold. To validate the methodology, we used IEEE-CIS fraud detection dataset for our experiment. Class imbalance and high dimensionality characteristics of dataset reduce the performance of model hence the data is preprocessed and trained. To evaluate the performance of the model, evaluation indicators such as precision, recall, f1-score, g-mean and Mathews Correlation Coefficient (MCC) are accomplished. The findings revealed that the performance of the proposed AE-XGB model is effective in handling imbalanced data and able to detect fraudulent transactions with 90.4% of recall and 90.5% of f1-score from incoming new transactions.

show abstract

Section: Discussionmentioning

confidence: 99%

Probabilistic XGBoost Threshold Classification with Autoencoder for Credit Card Fraud Detection

Prabha,

Priscilla

2023

IJRITCC

View full text Add to dashboard Cite

show abstract

“…Priscilla [11] et al proposed a two-stage feature selection method using mutual in-formation in the first stage and recursive feature elimination (RFE) in the second stage to eliminate redundant features. Pashaei [12] et al used minimum redundancy maximum relevance (mRMR) as a first-level filter and then introduced simulated annealing and crossover operators into a binary arithmetic optimization algorithm to select the mini-mum set of informative genes.…”

Section: Related Workmentioning

confidence: 99%

Feature Interaction Based Feature Selection Algorithm for In-trusion Detection

2024

AJCIS

View full text Add to dashboard Cite

Fog computing facilitates the placement of data at the network's edge for processing, which effectively reduces energy consumption and enhances efficiency. However, the limited resources inherent in fog computing render it vulnerable to extensive volumes of high-dimensional anomalous traffic. This study proposes a novel feature selection algorithm called filtered interaction maximum relevance minimum redundancy, which incorporates feature interaction to enable effective intrusion detection in fog computing. Through feature selection, the algorithm downscales the high-dimensional data captured in the fog nodes to reduce redundant features. The experimental results show that the parsimonious feature set obtained using the algorithm in this paper improves the classification accuracy while reducing the execution time compared to the original dataset.

show abstract

“…The identification of irrelevant or redundant features is a challenging task. Feature selection methods are used to choose a relevant subset of parameters that is sufficient to symbolize the original data [24]. The model with a reduced feature set will perform better at the time complexity.…”

Section: Literature Reviewmentioning

confidence: 99%

Optimal feature selection for cricket talent identification

2023

IJATEE

View full text Add to dashboard Cite

Cricket talent identification (TiD) is a methodical process that aims to find the young athletes possessing a potential to excel in the cricket sport at an early age. The sports scientists have identified a set of twenty-eight parameters that determine the cricket TiD. In order to realize the objective of computational efficiency by reducing the feature set, we perform an optimal feature selection for cricket TiD using nine different feature selection techniques Viz. mutual information, information gain ratio, correlation, chi square, univariate root mean square error (RMSE), receiver operating characteristic (ROC) with decision tree classifier, reliefF, boruta and oneR. The individual results obtained from the feature selection techniques are provided along with the individual ranking. We aggregate the results using two different rank aggregation techniques namely average ranking aggregation and majority vote ranking aggregation. The aggregation resultsshow a significant agreement between the two schemes. Fourteen out of twenty-eight features are selected using a threshold of 0.52-the value selected on recommendation of four different domain experts. 71.4% of the selected features are sport-centric and only 28.6% of the selected features are from the cognitive ability category. To the best of our knowledge, this is first such attempt to identify the talent in cricket using this methodology.

show abstract

A two-phase feature selection technique using mutual information and XGB-RFE for credit card fraud detection

Cited by 14 publications

References 32 publications

Probabilistic XGBoost Threshold Classification with Autoencoder for Credit Card Fraud Detection

Probabilistic XGBoost Threshold Classification with Autoencoder for Credit Card Fraud Detection

Feature Interaction Based Feature Selection Algorithm for In-trusion Detection

Optimal feature selection for cricket talent identification

Contact Info

Product

Resources

About