Recursive Feature Elimination Method Research Articles

Improving the accuracy of mangrove forest mapping using sentinel image and developed classification algorithm based on machine learning

ABSTRACT The national mangrove rehabilitation required mangrove spatial information with very high accuracy. This study successfully developed the classification algorithm measurable for mapping mangrove forest change based on machine learning. We also modified the recursive feature elimination (RFE) method to obtain the most optimal feature importance (FI). The results showed that the developed classification algorithm was built based on the five most important features including digital elevation model (DEM), near-infrared (NIR), normalized difference moisture index (NDMI), normalized difference water index (NDWI), and distances from brackish water river (DBWR). The developed algorithm increased overall and kappa accuracy by 0.48% and 0.01, respectively.

Comprehensive evaluation and performance analysis of machine learning in heart disease prediction

Heart disease is a leading cause of mortality on a global scale. Accurately predicting cardiovascular disease poses a significant challenge within clinical data analysis. The present study introduces a prediction model that utilizes various combinations of information and employs multiple established classification approaches. The proposed technique combines the genetic algorithm (GA) and the recursive feature elimination method (RFEM) to select relevant features, thus enhancing the model’s robustness. Techniques like the under sampling clustering oversampling method (USCOM) address the issue of data imbalance, thereby improving the model’s predictive capabilities. The classification challenge employs a multilayer deep convolutional neural network (MLDCNN), trained using the adaptive elephant herd optimization method (AEHOM). The proposed machine learning-based heart disease prediction method (ML-HDPM) demonstrates outstanding performance across various crucial evaluation parameters, as indicated by its comprehensive assessment. During the training process, the ML-HDPM model exhibits a high level of performance, achieving an accuracy rate of 95.5% and a precision rate of 94.8%. The system’s sensitivity (recall) performs with a high accuracy rate of 96.2%, while the F-score highlights its well-balanced performance, measuring 91.5%. It is worth noting that the specificity of ML-HDPM is recorded at a remarkable 89.7%. The findings underscore the potential of ML-HDPM to transform the prediction of heart disease and aid healthcare practitioners in providing precise diagnoses, exerting a substantial influence on patient care outcomes.

Feature selection using non-parametric correlations and important features on recursive feature elimination for stock price prediction

Stock price prediction using machine learning is a rapidly growing area of research. However, the large number of features that can be used can complicate the learning process. The feature selection method that can be used to overcome this problem is recursive feature elimination. Standard recursive feature elimination carries the risk of producing inaccurate algorithms because the top-ranked features are not necessarily the most important features. This research proposes a feature selection method that combines important features and nonparametric correlation in recursive feature elimination for stock price prediction. The data features used are technical indicators and stock price history. The recursive feature elimination method is modified with important features and nonparametric correlation features. The strategy for combining important features and non-parametric features is average weight, 25:75% weight, 75:25% weight, maximum weight, and minimum weight. The performance evaluation results show that the proposed feature selection method succeeded in obtaining small error values. The proposed method for predicting PT Bank Rakyat Indonesia Tbk (BBRI) stock prices obtains mean squared error, root mean square error, mean absolute error, and mean absolute percentage error evaluation values of 0.0000336, 0.00577, 0.00459, and 1.78%, respectively. This shows that recursive feature elimination with feature selection that combines important features and non-parametric correlation works better than the original recursive feature elimination at predicting stock prices.

Estimates of Crop Yield Anomalies for 2022 in Ukraine Based on Copernicus Sentinel-1, Sentinel-3 Satellite Data, and ERA-5 Agrometeorological Indicators.

The study explores the feasibility of adapting the EOStat crop monitoring system, originally designed for monitoring crop growth conditions in Poland, to fulfill the requirements of a similar system in Ukraine. The system utilizes satellite data and agrometeorological information provided by the Copernicus program, which offers these resources free of charge. To predict crop yields, the system uses several factors, such as vegetation condition indices obtained from Sentinel-3 Ocean and Land Color Instrument (OLCI) optical and Sea and Land Surface Temperature Radiometer (SLSTR). It also incorporates climate information, including air temperature, total precipitation, surface radiation, and soil moisture. To identify the best predictors for each administrative unit, the study utilizes a recursive feature elimination method and employs the Extreme Gradient Boosting regressor, a machine learning algorithm, to forecast crop yields. The analysis indicates a noticeable decrease in crop losses in 2022 in certain regions of Ukraine, compared to the previous year (2021) and the 5-year average (2017-2021), specifically for winter crops and maize. Considering the reduction in yield, it is estimated that the decline in production of winter crops in 2022 was up to 20%, while for maize, it was up to 50% compared to the decline in production.

An Analysis of Machine Learning for Detecting Depression, Anxiety, and Stress of Recovered COVID-19 Patients

Objectives: This study explores different machine learning models (KNN: k-nearest neighbor, MLP: Multilayer Perceptron, SVM: Support Vector Machine) to identify the optimal model for accurate and rapid mental health detection among the recovered COVID-19 patients. Other techniques are also investigated, such as feature selection (Recursive Feature Elimination (RFE) and Extra Trees (ET) methods) and hyper-parameter tuning, to achieve a system that could effectively and quickly indicate mental health. Method/Analysis: To achieve the objectives, the study employs a dataset collected from recovered COVID-19 patients, encompassing information related to depression, anxiety, and stress. Machine learning models are utilized in the analysis. Additionally, feature selection methods and hyper-parameter tuning techniques are explored to enhance the model’s predictive capabilities. The performance of each model is assessed based on accuracy metrics. Findings: The experimental results show that SVM is the most suitable model for accurately predicting an individual’s mental health among recovered COVID-19 patients (accuracy ≥ 0.984). Furthermore, the ET method is more effective than the RFE method for feature selection in the anxiety and stress datasets. Novelty/Improvement:The study lies in the understanding of predictive modeling for mental health and provides insights into the choice of models and techniques for accurate and early detection. Doi: 10.28991/HEF-2024-05-01-01 Full Text: PDF

Integrating Meteorological and Remote Sensing Data to Simulate Cropland Nocturnal Evapotranspiration Using Machine Learning

Evapotranspiration (ET) represents a significant component of the global water flux cycle, yet nocturnal evapotranspiration (ETn) is often neglected, leading to underestimation of global evapotranspiration. As for cropland, accurate modeling of ETn is essential for rational water management and is important for sustainable agriculture development. We used random forest (RF) to simulate ETn at 16 globally distributed cropland eddy covariance flux sites along with remote sensing and meteorological factors. The recursive feature elimination method was used to remove unimportant variables. We also simulated the ETn of C3 and C4 crops separately. The trained RF resulted in a determination coefficient (R2) (root mean square error (RMSE)) of 0.82 (7.30 W m−2) on the testing dataset. C3 and C4 crops on the testing dataset resulted in an R2 (RMSE) of 0.86 (5.59 W m−2) and 0.55 (4.86 W m−2) for the two types of crops. We also showed that net radiation is the dominant factor in regulating ETn, followed by 2 m horizontal wind speed and vapor pressure deficit (VPD), and these three meteorological factors showed a significant positive correlation with ETn. This research demonstrates that RF can simulate ETn from crops economically and accurately, providing a methodological basis for improving global ETn simulations.

Bioinformatics Analysis and Experimental Validation of Mitochondrial Autophagy Genes in Knee Osteoarthritis.

Mitochondrial autophagy is closely related to the pathogenesis of osteoarthritis, In order to explore the role of mitochondrial autophagy related genes in knee osteoarthritis (KOA) and its molecular mechanism. KOA-related transcriptome data were extracted from the Gene Expression Omnibus (GEO) database. Differentially expressed mitochondrial autophagy gene (DEMGs) were screened in patients with KOA by differential expression analysis. The STRING website was used to construct a protein-protein interaction (PPI) network among DEMGs. Molecular complex detection (MCODE) method in Cytoscape software was performed to identify hub DEMGs. Support vector machine recursive feature elimination (SVM-RFE) method was used to construct the hub DEMG diagnosis model. Genes with diagnostic value were identified as biomarkers by plotting receiver operating characteristic (ROC) curves and Expression validation. CIBERSORT algorithm was used to calculate the proportion of 22 immune cells in each sample in the GSE114007 dataset. Finally, biomarker expression was verified by qPCR. A total of 15 DEMGs were obtained and enrichment analyses showed that these DEMG strains were mainly enriched in the mitophagy-animal, shigellosis, autophagy-animal and FoxO signal pathways. The PPI network unveiled 13 DEMGs with interactions. In addition, 8 hub DEMGs (ULK1, CALCOCO2, MAP1LC3B, BNIP3L, GABARAPL1, BNIP3, FKBP8 and FOXO3) were obtained for KOA. And 5 model DEMGs (BNIP3L, BNIP3, MAP1LC3B, ULK1 and FOXO3) were screened. The ROC curves revealed that BNIP3 and FOXO3 has strong diagnostic value in these models of DEMG. Immune-infiltration and correlation analysis showed that BNIP3 and FOXO3 were significantly correlated with three different immune cells, including primary B cells, M0 macrophage and M2 macrophage. The cartilage tissue samples qPCR verification results show that FOXO3 and BNIP3 were all down-regulated in KOA (p < 0.01), and the validation results are consistent with the above analysis. BNIP3 and FOXO3 have been identified as biomarkers for the diagnosis of KOA, which might supply a new insight for the pathogenesis and treatment of KOA.

Application of radiomics model based on ultrasound image features in the prediction of carpal tunnel syndrome severity.

The aim of our study is to develop and validate a radiomics model based on ultrasound image features for predicting carpal tunnel syndrome (CTS) severity. This retrospective study included 237 CTS hands (106 for mild symptom, 68 for moderate symptom and 63 for severe symptom). There were no statistically significant differences among the three groups in terms of age, gender, race, etc. The data set was randomly divided into a training set and a test set in a ratio of 7:3. Firstly, a senior musculoskeletal ultrasound expert measures the cross-sectional area of median nerve (MN) at the scaphoid-pisiform level. Subsequently, a recursive feature elimination (RFE) method was used to identify the most discriminative radiomic features of each MN at the entrance of the carpal tunnel. Eventually, a random forest model was employed to classify the selected features for prediction. To evaluate the performance of the model, the confusion matrix, receiver operating characteristic (ROC) curves, and F1 values were calculated and plotted correspondingly. The prediction capability of the radiomics model was significantly better than that of ultrasound measurements when 10 robust features were selected. The training set performed perfect classification with 100% accuracy for all participants, while the testing set performed accurate classification of severity for 76.39% of participants with F1 values of 80.00, 63.40, and 84.80 for predicting mild, moderate, and severe CTS, respectively. Comparably, the F1 values for mild, moderate, and severe CTS predicted based on the MN cross-sectional area were 76.46, 57.78, and 64.00, respectively.. This radiomics model based on ultrasound images has certain value in distinguishing the severity of CTS, and was slightly superior to using only MN cross-sectional area for judgment. Although its diagnostic efficacy was still inferior to that of neuroelectrophysiology. However, this method was non-invasive and did not require additional costs, and could provide additional information for clinical physicians to develop diagnosis and treatment plans.

Prediction of the Fatal Acute Complications of Myocardial Infarction via Machine Learning Algorithms.

Myocardial infarction (MI) is a major cause of death, particularly during the first year. The avoidance of potentially fatal outcomes requires expeditious preventative steps. Machine learning (ML) is a subfield of artificial intelligence science that detects the underlying patterns of available big data for modeling them. This study aimed to establish an ML model with numerous features to predict the fatal complications of MI during the first 72 hours of hospital admission. We applied an MI complications database that contains the demographic and clinical records of patients during the 3 days of admission based on 2 output classes: dead due to the known complications of MI and alive. We utilized the recursive feature elimination (RFE) method to apply feature selection. Thus, after applying this method, we reduced the number of features to 50. The performance of 4 common ML classifier algorithms, namely logistic regression, support vector machine, random forest, and extreme gradient boosting (XGBoost), was evaluated using 8 classification metrics (sensitivity, specificity, precision, false-positive rate, false-negative rate, accuracy, F1-score, and AUC). In this study of 1699 patients with confirmed MI, 15.94% experienced fatal complications, and the rest remained alive. The XGBoost model achieved more desirable results based on the accuracy and F1-score metrics and distinguished patients with fatal complications from surviving ones (AUC=78.65%, sensitivity=94.35%, accuracy=91.47%, and F1-score=95.14%). Cardiogenic shock was the most significant feature influencing the prediction of the XGBoost algorithm. XGBoost algorithms can be a promising model for predicting fatal complications following MI.

A Wearable Inertial Sensor Approach for Locomotion and Localization Recognition on Physical Activity.

Advancements in sensing technology have expanded the capabilities of both wearable devices and smartphones, which are now commonly equipped with inertial sensors such as accelerometers and gyroscopes. Initially, these sensors were used for device feature advancement, but now, they can be used for a variety of applications. Human activity recognition (HAR) is an interesting research area that can be used for many applications like health monitoring, sports, fitness, medical purposes, etc. In this research, we designed an advanced system that recognizes different human locomotion and localization activities. The data were collected from raw sensors that contain noise. In the first step, we detail our noise removal process, which employs a Chebyshev type 1 filter to clean the raw sensor data, and then the signal is segmented by utilizing Hamming windows. After that, features were extracted for different sensors. To select the best feature for the system, the recursive feature elimination method was used. We then used SMOTE data augmentation techniques to solve the imbalanced nature of the Extrasensory dataset. Finally, the augmented and balanced data were sent to a long short-term memory (LSTM) deep learning classifier for classification. The datasets used in this research were Real-World Har, Real-Life Har, and Extrasensory. The presented system achieved 89% for Real-Life Har, 85% for Real-World Har, and 95% for the Extrasensory dataset. The proposed system outperforms the available state-of-the-art methods.

Multisequence MRI-based radiomics signature as potential biomarkers for differentiating KRAS mutations in non-small cell lung cancer with brain metastases

BackgroundKirsten rat sarcoma virus (KRAS) has evolved from a genotype with predictive value to a therapeutic target recently. The study aimed to establish non-invasive radiomics models based on MRI to discriminate KRAS from epidermal growth factor receptor (EGFR) or anaplastic lymphoma kinase (ALK) mutations in lung cancer patients with brain metastases (BM), then further explore the optimal sequence for prediction. MethodsThis retrospective study involved 317 patients (218 patients in training cohort and 99 patients in testing cohort) who had confirmed of KRAS, EGFR or ALK mutations. Radiomics features were separately extracted from T2WI, T2 fluid-attenuated inversion recovery (T2-FLAIR), diffusion weighted imaging (DWI) and contrast-enhanced T1-weighted imaging (T1-CE) sequences. The maximal information coefficient and recursive feature elimination method were used to select informative features. Then we built four radiomics models for differentiating KRAS from EGFR or ALK using random forest classifier. ROC curves were used to validate the capability of the models. ResultsThe four radiomics models for discriminating KRAS from EGFR all worked well, especially DWI and T2WI models (AUCs: 0.942, 0.942 in training cohort, 0.949, 0.954 in testing cohort). When KRAS compared to ALK, DWI and T2-FLAIR models showed excellent performance in two cohorts (AUCs: 0.947, 0.917 in training cohort, 0.850, 0.824 in testing cohort). ConclusionsRadiomics classifiers integrating MRI have potential to discriminate KRAS from EGFR or ALK, which are helpful to guide treatment and facilitate the discovery of new approaches capable of achieving this long-sought goal of cure in lung cancer patients with KRAS.

Sparse Deep Neural Network for Encoding and Decoding the Structural Connectome.

Brain state classification by applying deep learning techniques on neuroimaging data has become a recent topic of research. However, unlike domains where the data is low dimensional or there are large number of available training samples, neuroimaging data is high dimensional and has few training samples. To tackle these issues, we present a sparse feedforward deep neural architecture for encoding and decoding the structural connectome of the human brain. We use a sparsely connected element-wise multiplication as the first hidden layer and a fixed transform layer as the output layer. The number of trainable parameters and the training time is significantly reduced compared to feedforward networks. We demonstrate superior performance of this architecture in encoding the structural connectome implicated in Alzheimer's disease (AD) and Parkinson's disease (PD) from DTI brain scans. For decoding, we propose recursive feature elimination (RFE) algorithm based on DeepLIFT, layer-wise relevance propagation (LRP), and Integrated Gradients (IG) algorithms to remove irrelevant features and thereby identify key biomarkers associated with AD and PD. We show that the proposed architecture reduces 45.1% and 47.1% of the trainable parameters compared to a feedforward DNN with an increase in accuracy by 2.6 % and 3.1% for cognitively normal (CN) vs AD and CN vs PD classification, respectively. We also show that the proposed RFE method leads to a further increase in accuracy by 2.1% and 4% for CN vs AD and CN vs PD classification, while removing approximately 90% to 95% irrelevant features. Furthermore, we argue that the biomarkers (i.e., key brain regions and connections) identified are consistent with previous literature. We show that relevancy score-based methods can yield high discriminative power and are suitable for brain decoding. We also show that the proposed approach led to a reduction in the number of trainable network parameters, an increase in classification accuracy, and a detection of brain connections and regions that were consistent with earlier studies.

Performance and explainability of feature selection-boosted tree-based classifiers for COVID-19 detection

In this paper, we evaluate the performance and analyze the explainability of machine learning models boosted by feature selection in predicting COVID-19-positive cases from self-reported information. In essence, this work describes a methodology to identify COVID-19 infections that considers the large amount of information collected by the University of Maryland Global COVID-19 Trends and Impact Survey (UMD-CTIS). More precisely, this methodology performs a feature selection stage based on the recursive feature elimination (RFE) method to reduce the number of input variables without compromising detection accuracy. A tree-based supervised machine learning model is then optimized with the selected features to detect COVID-19-active cases. In contrast to previous approaches that use a limited set of selected symptoms, the proposed approach builds the detection engine considering a broad range of features including self-reported symptoms, local community information, vaccination acceptance, and isolation measures, among others. To implement the methodology, three different supervised classifiers were used: random forests (RF), light gradient boosting (LGB), and extreme gradient boosting (XGB). Based on data collected from the UMD-CTIS, we evaluated the detection performance of the methodology for four countries (Brazil, Canada, Japan, and South Africa) and two periods (2020 and 2021). The proposed approach was assessed in terms of various quality metrics: F1-score, sensitivity, specificity, precision, receiver operating characteristic (ROC), and area under the ROC curve (AUC). This work also shows the normalized daily incidence curves obtained by the proposed approach for the four countries. Finally, we perform an explainability analysis using Shapley values and feature importance to determine the relevance of each feature and the corresponding contribution for each country and each country/year.

Comorbidities Incorporated to Improve Prediction for Prevalent Mild Cognitive Impairment and Alzheimer's Disease in the HABS-HD Study.

Blood biomarkers have the potential to transform Alzheimer's disease (AD) diagnosis and monitoring, yet their integration with common medical comorbidities remains insufficiently explored. This study aims to enhance blood biomarkers' sensitivity, specificity, and predictive performance by incorporating comorbidities. We assess this integration's efficacy in diagnostic classification using machine learning, hypothesizing that it can identify a confident set of predictive features. We analyzed data from 1,705 participants in the Health and Aging Brain Study-Health Disparities, including 116 AD patients, 261 with mild cognitive impairment, and 1,328 cognitively normal controls. Blood samples were assayed using electrochemiluminescence and single molecule array technology, alongside comorbidity data gathered through clinical interviews and medical records. We visually explored blood biomarker and comorbidity characteristics, developed a Feature Importance and SVM-based Leave-One-Out Recursive Feature Elimination (FI-SVM-RFE-LOO) method to optimize feature selection, and compared four models: Biomarker Only, Comorbidity Only, Biomarker and Comorbidity, and Feature-Selected Biomarker and Comorbidity. The combination model incorporating 17 blood biomarkers and 12 comorbidity variables outperformed single-modal models, with NPV12 at 92.78%, AUC at 67.59%, and Sensitivity at 65.70%. Feature selection led to 22 chosen features, resulting in the highest performance, with NPV12 at 93.76%, AUC at 69.22%, and Sensitivity at 70.69%. Additionally, interpretative machine learning highlighted factors contributing to improved prediction performance. In conclusion, combining feature-selected biomarkers and comorbidities enhances prediction performance, while feature selection optimizes their integration. These findings hold promise for understanding AD pathophysiology and advancing preventive treatments.

Big data analytics for relative humidity time series forecasting based on the LSTM network and ELM

Accurate and reliable relative humidity forecasting is important when evaluating the impacts of climate change on humans and ecosystems. However, the complex interactions among geophysical parameters are challenging and may result in inaccurate weather forecasting. This study combines long short-term memory (LSTM) and extreme learning machines (ELM) to create a hybrid model-based forecasting technique to predict relative humidity to improve the accuracy of forecasts. Detailed experiments with univariate and multivariate problems were conducted, and the results show that LSTM-ELM and ELM-LSTM have the lowest MAE and RMSE results compared to stand-alone LSTM and ELM for the univariate problem. In addition, LSTM-ELM and ELM-LSTM result in lower computation time than stand-alone LSTM. The experiment results demonstrate that the proposed hybrid models outperform the comparative methods in relative humidity forecasting. We employed the recursive feature elimination (RFE) method and showed that dewpoint temperature, temperature, and wind speed are the factors that most affect relative humidity. A higher dewpoint temperature indicates more air moisture, equating to high relative humidity. Humidity levels also rise as the temperature rises.

A novel multi-layer stacking ensemble wind power prediction model under Tensorflow deep learning framework considering feature enhancement and data hierarchy processing

Wind power prediction is crucial for energy production, but due to the complicated data characteristics of wind farms, it's difficult to accurately predict wind power output and it is challenging for a single prediction model to properly handle multi-featured time-series data sets. To solve the above problem, the density-based spatial clustering of applications with noise (DBSCAN) method was used to find outliers of data sets, and the recursive feature elimination (RFE) method was used to screen out main features according to importance. Sequentially a multi-layer stacking ensemble learning prediction model was constructed based on data hierarchy processing and feature enhancement techniques. Seven groups of ablation experiments were set to verify the model's effectiveness with the annual time-series data. The experimental results show that the DBSCAN method can effectively find outliers in wind data sets and improve the prediction accuracy and the RFE method can significantly reduce the computing time and improves the generalization ability and prediction accuracy. The multi-layer stacking ensemble model can slow down the overfitting trend of a single prediction model, effectively avoid the model from falling into the local search state, and obtain a strong learner with better generalization ability.

Efficient Prediction of Seasonal Infectious Diseases Using Hybrid Machine Learning Algorithms with Feature Selection Techniques

Early seasonal disease risk identification is the most challenging task in the medical industry. The capacity to detect patients at risk of improvement throughout their hospital stay is critical for effective patient allocation and care among patients with seasonal diseases. Patient risk factor prediction is the most important issue in reducing victim mortality. Machine learning plays a vital role in identifying the risk level of patients. In this research, we consider four seasonal diseases such as dengue, malaria, typhoid, and pneumonia. In this research, the researcher finds the dangerous symptoms of victims based on their age group. The real-time dataset was used in this study. The dataset for this study was gathered from hospitals in the Madurai district between 2019 and 2020. Feature selection is the prime element of patient risk recognition. It is used to pick the most accurate attributes for prediction. The dataset is divided into 70% training data and 30% testing data. This research proposes the feature selection method Boruta-XGBoost for improving accuracy. In this research, we discuss various attribute selection algorithms, including the Boruta algorithm, the XGBoost algorithm, the recursive feature elimination method (RFE), and the PRF-BXGBoost (Patient Risk Factor-Boruta-XGBoost) algorithm. The proposed method provides greater accuracy when compared to other variable selection methods. The time complexity of the proposed method is low when compared to other algorithms.

A Unified Multi-Class Feature Selection Framework for Microarray Data.

In feature selection research, simultaneous multi-class feature selection technologies are popular because they simultaneously select informative features for all classes. Recursive feature elimination (RFE) methods are state-of-the-art binary feature selection algorithms. However, extending existing RFE algorithms to multi-class tasks may increase the computational cost and lead to performance degradation. With this motivation, we introduce a unified multi-class feature selection (UFS) framework for randomization-based neural networks to address these challenges. First, we propose a new multi-class feature ranking criterion using the output weights of neural networks. The heuristic underlying this criterion is that "the importance of a feature should be related to the magnitude of the output weights of a neural network". Subsequently, the UFS framework utilizes the original features to construct a training model based on a randomization-based neural network, ranks these features by the criterion of the norm of the output weights, and recursively removes a feature with the lowest ranking score. Extensive experiments on 15 real-world datasets suggest that our proposed framework outperforms state-of-the-art algorithms. The code of UFS is available at https://github.com/SVMrelated/UFS.git.

Feature selection for Turkish Crowdfunding projects with using filtering and wrapping methods

Crowdfunding (CF) platforms host an increasing number of projects, where financial support from backers plays a vital role in project realization. Unfortunately, CF projects have experienced a downward trend in success rates. To address this issue, it is crucial to identify the factors that influence success by analyzing project characteristics. In our study, we collected project data from Turkey's Fongogo CF platform, performed feature selection, and rigorously tested the results. We employed various methods such as Pearson correlation, mutual information statistics, chi-square, Fisher's score from filtering methods, and recursive feature elimination from wrapper methods to understand feature relationships. We proposed a cross-validated recursive feature elimination method for feature selection. The identified success factors were classified using diverse machine learning algorithms, with the Gradient Boosting algorithm achieving the highest result of 84.28%. The results obtained with wrapper methods highlight the potential of utilizing features in decision support processes to enhance CF success.

An Intrusion Detection Method Based on Attention Mechanism to Improve CNN-BiLSTM Model

Abstract Security of computer information can be improved with the use of a network intrusion detection system. Since the network environment is becoming more complex, more and more new methods of attacking the network have emerged, making the original intrusion detection methods ineffective. Increased network activity also causes intrusion detection systems to identify errors more frequently. We suggest a new intrusion detection technique in this research that combines a Convolutional Neural Network (CNN) model with a Bi-directional Long Short-term Memory Network (BiLSTM) model for adding attention mechanisms. We distinguish our model from existing methods in three ways. First, we use the NCR-SMOTE algorithm to resample the dataset. Secondly, we use recursive feature elimination method based on extreme random tree to select features. Thirdly, we improve the profitability and accuracy of predictions by adding attention mechanism to CNN-BiLSTM. This experiment uses UNSW-UB15 dataset composed of real traffic, and the accuracy rate of multi-classification is 84.5$\%$; the accuracy rate of multi-classification in CSE-IC-IDS2018 dataset reached 98.3$\%$.