Gain Ratio Feature Selection Research Articles

A novel credit scoring system in financial institutions using artificial intelligence technology

<p>In order to evaluate a person’s or a company’s creditworthiness, financial institutions must use credit scoring. Traditional credit scoring algorithms frequently rely on manual and rule-based methods, which can be tedious and inaccurate. Recent developments in artificial intelligence (AI) technology have opened up possibilities for creating more reliable and effective credit rating systems. The data are pre-processed, including scaling using the 0–1 normalization method and resolving missing values by imputation. Information gain (IG), gain ratio (GR), and chi-square are three feature selection methodologies covered in the study. While GR normalizes IG by dividing it by the total entropy of the feature, IG quantifies the reduction in total entropy by adding a new feature. Based on chi-squared statistics, the most vital traits are determined using chi-square. This research employs different ML models to develop a hybrid model for credit score prediction. The ML algorithms support vector machine (SVM), neural networks (NNs), decision trees (DTs), random forest (RF), and logistic regression (LR) classifiers are employed here for experiments along with IG, GR, and chi-square feature selection methodologies for credit prediction over Australian and German datasets. The study offers an understanding of the decision-making process for informative characteristics and the functionality of machine learning (ML) in credit prediction tasks. The empirical analysis shows that in the case of the German dataset, the DT with GR feature selection and hyperparameter optimization outperforms SVM and NN with an accuracy of 99.78%. For the Australian dataset, SVM with GR feature selection outperforms NN and DT with an accuracy of 99.98%.</p>

Performance Analysis of LVQ 1 Using Feature Selection Gain Ratio for Sex Classification in Forensic Anthropology

One approach to handling large of data dimensions is feature selection. Effective feature selection techniques produce the essential features and can improve classification algorithms. The accuracy performance results can measure the accuracy of the method used in the classification process. This research uses the Learning Vector Quantization (LVQ) 1 method combined with Gain Ratio feature selection. The data used is male and female skull bone measurement data totaling 2524. The highest accuracy results are obtained by LVQ 1, which uses a Gain Ratio with a threshold of 0.01 with a learning rate = 0.1, which is 92.01%, and the default threshold weka(-1.7976931348623157E308) with a learning rate = 0.1, which is 92.19%. In comparison, previous research that did not use gain ratio or that did not use GR only had the best results of 91.39% with a learning rate = 0.1, 0.4, 0.7, 0.9. This shows that LVQ 1 using the Gain Ratio can be recommended to improve the performance of the Skull dataset compared to LVQ 1 without Gain Ratio.

A comparative study of metaheuristic-based machine learning classifiers using non-parametric tests for the detection of COPD severity grade

In this paper, a comparative study between five different Machine Learning (ML) classifiers has been performed for the detection of Chronic Obstructive Pulmonary Disease (COPD) severity grade. There are various existing studies that provide various computer-aided frameworks for the diagnosis of COPD. However, such existing approaches have certain limitations such as lower performance, imbalanced data, and missing data. Also, no statistical tests were performed that could justify the validity of the proposed techniques. Hence, to provide the best model for the detection of COPD severity grade, a conceptualized framework has been provided. The performance of ML classifiers has been optimized by utilizing the Correlation-based Elephant search algorithm (CFS+ESA) and InformationGain and GainRatio feature selection techniques (FS). Experimental work carried out on the COPD patient dataset has been validated using statistical tests. The Wilcoxon test and Friedman test were utilized to compare the FS techniques and ML techniques respectively. The performance comparison has been carried out across different training-testing criteria namely, 10-fold cross-validation, 70%-30%, 75%-25%, and 80%-20%. The results from Wilcoxon tests showed that CFS+ESA has given the best results across all the training-testing criteria. Similarly, the results from Friedman’s tests demonstrated that the CFS+ESA-based Logistic Regression classifier trained using 10-fold cross-validation has outperformed the rest combinations of classifiers and training-testing criteria by achieving the highest rank.

Advanced hybrid ensemble gain ratio feature selection model using machine learning for enhanced disease risk prediction

Currently, there is an increased need for employing machine learning (ML) and data mining in the healthcare system domain, applications of which play a pivotal role in providing beneficial knowledge to society by utilizing the available data. There are disease risk prediction models that either do not use feature selection or use traditional feature selection models—filter, wrapper, and embedded—which have limitations such as, not using ML, using single evaluation metric, and working on imbalanced datasets. Apart from this, there is a great scope for enhancement in the prediction performance. To address these issues, an advanced hybrid ensemble gain ratio feature selection (AHEG-FS) model is proposed, which consists of four major feature selection techniques: an ensemble feature selection, a gain ratio feature selection, a backward feature elimination, and an area under the curve (AUC)—an additional evaluation metric of the novel feature reduction along with accuracy, for robust feature selection aimed at effective disease risk prediction. The subsets of important and highly ranked features are obtained with the first two techniques. Then, the proposed model is aligned with nine ML algorithms. Additionally, in the third and fourth techniques of the proposed model, the AUCs are evaluated for the aforementioned ML algorithms and the backward feature elimination is applied to remove the redundant features, resulting in the acquisition of the best subsets of highly contributing features that produce the highest precision results. Thus, four benchmarked heart disease datasets—Cleveland, Hungarian, Statlog, and Switzerland of the University of California, Irvine ML repository—are used, and the results are encouraging. The highest AUC and accuracy of 99.00% and 95.47%, respectively, are achieved, with 46.15% of features reduced. A 6.18% higher accuracy than recent studies was achieved with convergent speed.

Cloud Computing-Based Framework for Breast Cancer Diagnosis Using Extreme Learning Machine.

Globally, breast cancer is one of the most significant causes of death among women. Early detection accompanied by prompt treatment can reduce the risk of death due to breast cancer. Currently, machine learning in cloud computing plays a pivotal role in disease diagnosis, but predominantly among the people living in remote areas where medical facilities are scarce. Diagnosis systems based on machine learning act as secondary readers and assist radiologists in the proper diagnosis of diseases, whereas cloud-based systems can support telehealth services and remote diagnostics. Techniques based on artificial neural networks (ANN) have attracted many researchers to explore their capability for disease diagnosis. Extreme learning machine (ELM) is one of the variants of ANN that has a huge potential for solving various classification problems. The framework proposed in this paper amalgamates three research domains: Firstly, ELM is applied for the diagnosis of breast cancer. Secondly, to eliminate insignificant features, the gain ratio feature selection method is employed. Lastly, a cloud computing-based system for remote diagnosis of breast cancer using ELM is proposed. The performance of the cloud-based ELM is compared with some state-of-the-art technologies for disease diagnosis. The results achieved on the Wisconsin Diagnostic Breast Cancer (WBCD) dataset indicate that the cloud-based ELM technique outperforms other results. The best performance results of ELM were found for both the standalone and cloud environments, which were compared. The important findings of the experimental results indicate that the accuracy achieved is 0.9868, the recall is 0.9130, the precision is 0.9054, and the F1-score is 0.8129.

Stress Detection via Keyboard Typing Behaviors by Using Smartphone Sensors and Machine Learning Techniques.

Stress is one of the biggest problems in modern society. It may not be possible for people to perceive if they are under high stress or not. It is important to detect stress early and unobtrusively. In this context, stress detection can be considered as a classification problem. In this study, it was investigated the effects of stress by using accelerometer and gyroscope sensor data of the writing behavior on a smartphone touchscreen panel. For this purpose, smartphone data including two states (stress and calm) were collected from 46 participants. The obtained sensor signals were divided into 5, 10 and 15s interval windows to create three different data sets and 112 different features were defined from the raw data. To obtain more effective feature subsets, these features were ranked by using Gain Ratio feature selection algorithm. Afterwards, writing behaviors were classified by C4.5 Decision Trees, Bayesian Networks and k-Nearest Neighbor methods. As a result of the experiments, 74.26%, 67.86%, and 87.56% accuracy classification results were obtained respectively.

Effects of Machine Learning Approach in Flow-Based Anomaly Detection on Software-Defined Networking

Recent advancements in software-defined networking (SDN) make it possible to overcome the management challenges of traditional networks by logically centralizing the control plane and decoupling it from the forwarding plane. Through a symmetric and centralized controller, SDN can prevent security breaches, but it can also bring in new threats and vulnerabilities. The central controller can be a single point of failure. Hence, flow-based anomaly detection system in OpenFlow Controller can secure SDN to a great extent. In this research, we investigated two different approaches of flow-based intrusion detection system in OpenFlow Controller. The first of which is based on machine-learning algorithm where NSL-KDD dataset with feature selection ensures the accuracy of 82% with random forest classifier using the gain ratio feature selection evaluator. In the later phase, the second approach is combined with a deep neural network (DNN)-based intrusion detection system based on gated recurrent unit-long short-term memory (GRU-LSTM) where we used a suitable ANOVA F-Test and recursive feature elimination selection method to boost classifier output and achieve an accuracy of 88%. Substantial experiments with comparative analysis clearly show that, deep learning would be a better choice for intrusion detection in OpenFlow Controller.

Stock daily return prediction using expanded features and feature selection

Stock market prediction is a very noisy problem and the use of any additional information to increase accuracy is necessary. In this paper, for the stock daily return prediction problem, the set of features is expanded to include indicators not only for the stock to be predicted itself but also a set of other stocks and currencies. Afterwards, different feature selection and classification methods are utilized for prediction. The daily close returns of the 3 most traded stocks (GARAN, THYAO, and ISCTR) in Borsa Istanbul (BIST) are predicted using indicators computed on those stocks, indicators for all the other stocks listed in the BIST100 index, and indicators on the dollar-gold prices. Twenty-five different indicators on daily stock prices are computed to form feature vectors for each trading day. These feature vectors are assigned class labels according to the daily close returns. Expanding the feature space with BIST100 stocks features results in a high dimensional feature space, with possibly noisy or irrelevant features. Therefore, feature selection methods are utilized to select the most informative features. In order to determine relevance scores of features, fast filter-based methods, gain ratio and relief, are used. Experiments are performed based on individual stock features, dollar-gold features (DG), BIST100 stock features (BIST100), and a combination of BIST100 and DG with and without feature selection. Using the gain ratio feature selection with a gradient boosting machine (GBM), the movements of GARAN stock were predicted with an accuracy of 0.599 and an F-measure of 0.614. For THYAO, the relief feature selection with the GBM gave an accuracy of 0.558, and for ISCTR, the gain ratio feature selection with logistic regression achieved an accuracy of 0.581. It was found that using BIST100 stock features boosts classification results for all stocks in terms of accuracy.

Wavelet-based energy features for diagnosis of melanoma from dermoscopic images

Melanoma causes majority of deaths related to skin cancer if not detected and treated at an early stage. It is considered as one of the dangerous types of cancer, since it quickly spreads to other parts. A novel approach has been proposed in this paper to detect Melanoma from dermoscopic images. Pre-processing is done to remove hair and noise in the image. Initial segmentation is carried out with watershed transform. This is followed by Maximal Similarity Region Merging process. After pre-processing and segmentation, wavelet-based energy features are extracted using daubechies (DB3), and reverse biorthogonal (RBIO3.3, RBIO3.5, and RBIO3.7) wavelet filters. 12 features are extracted using four wavelet filters. Using the Gain Ratio feature selection method, the most discriminative six features are selected for the classification purpose. Then classification has been done by using K-nearest neighbour, support vector machine, random forest, and Naive Bayes classifier. The highest sensitivity of 97.5% is achieved in the case of support vector machine.

An Ensemble Model for Classification of Attacks with Feature Selection based on KDD99 and NSL-KDD Data Set

Information security is extremely critical issues for every organization to protect information from unauthorized access. Intrusion detection system has one of the important roles to prevent data or information from malicious behaviours. Basically Intrusion detection system is a classifier that can classify the data as normal or attacks. In this research paper, we have proposed ANN-Bayesian Net-GR technique that means ensemble of Artificial Neural Network (ANN) and Bayesian Net with Gain Ratio (GR) feature selection technique. We have applied various individual classification techniques and its ensemble model on KDD99 and NSL-KDD data set to check the robustness of model. Due to irrelevant features in data set, also applied Gain Ratio feature selection technique on best model. Finally our proposed model produces highest accuracy compare to others.

Ensemble Learning Model for Diabetes Classification

This paper proposed data mining techniques to improve efficiency and reliability in diabetes classification. The real data set collected from Sawanpracharak Regional Hospital, Thailand, was fist analyzed by using gain-ratio feature selection techniques. Three well known algorithms; naïve bayes, k-nearest neighbors and decision tree, were used to construct classification models on the selected features. Then, the popular ensemble learning; bagging and boosting were applied using the three base classifiers. The results revealed that the best model with the highest accuracy was bagging with base classifier decision tree algorithm (95.312%). The experiments also showed that ensemble classifier models performed better than the base classifiers alone.

Two lipids based on lipidomics as novel biomarkers for early detection of squamous cell lung cancer.

11114 Background: Lipids play roles in membrane structure, energy storage, and signal transduction as well as lung cancer. Lipidomics, a new technology aims to measure all the lipids in a cell, has not been applied to diagnostic test development for a variety of cancer types. Here, we adopt lipidomics as a means to identify plasma lipid markers for the early detection of lung cancer and complement CT-based methods for lung cancer screening. Methods: Using mass spectrometry, we profiled 390 individual lipids in a training discovery cohort comprised of cohorts that were either at “high-risk” for lung cancer (n=22) and squamous cell carcinoma at early stages (n=22). Cases had a minimum of two years clinical follow-up and were matched in terms of race, sex, age and smoking status. Gain ratio feature selection and local weighted classification model were employed to find the best training classifier, which was further validated against an additional cohort, including high-risk individuals (n= 20) and squamous cell carcinoma patients (n=17). Results: In the training discovery stage, we found 20 distinct lipids that were significantly distributed between high-risk and cases of squamous cell carcinoma. We further defined a two lipid marker panel had a training accuracy at 95.5% sensitivity, 90.9% specificity and 95.2% AUC (Area under ROC curve). The validation accuracy against the additional cohort is 100.0% sensitivity, 90.0% specificity and 99.0% AUC (Table). The power for sample size we used in both discovery training and validation stages were over 90%. Conclusions: Using lipidomics we identified two lipid markers capable of discerning cases of squamous cell carcinoma from individuals at high risk for lung cancer, with a high sensitivity, specificity and accuracy. The markers maybe further developed as a quick, safe blood test for early diagnosis of squamous cell lung cancer and reduce unnecessary follow-up imaging or invasive procedures. [Table: see text]