A review on feature selection and feature extraction for text classification

Feature engineering is treated as a crucial step in any intelligent system pipeline and practitioners often spend 70–80% of their time in this phase before modeling. Especially, in machine learning (ML), the performance of the classification systems deteriorates as irrelevant features are added, even when the features presented contain enough information about the problem. ML researchers often use feature selection and feature extraction techniques individually to filter out the redundant information from the data, but would not get desired results or performance due to choosing a single method only. In this article, we take an attempt to combine both feature selection and feature extraction methods that give an improvement in the prediction performance than using a single feature selection or feature extraction method. In the application of the proposed method, we consider the Indian liver disease patient dataset to classify liver disease using machine learning classifiers. In addition, ensemble classifiers also take into account reproducing and robustifying classification accuracy. The performance is evaluated and compared with different metrics such as accuracy, precision, recall, and F1-score. Using the proposed method, the KNN classifier can diagnosis 76.03% with a precision of 76.69%, recall 96.23%, and 85.36% F1-score which is much better than the individual feature selection and feature extraction approach. The proposed method can also be extended to medical datasets other than liver disease so that these types of deadly diseases can be correctly diagnosed for the betterment of humanity.

Dimensionality reduction as a preprocessing step to machine learning is effective in removing irrelevant and redundant data, increasing learning accuracy, and improving result comprehensibility. However, the recent increase of dimensionality of data poses a severe challenge to many existing feature selection and feature extraction methods with respect to efficiency and effectiveness. In the field of machine learning and pattern recognition, dimensionality reduction is important area, where many approaches have been proposed. In this paper, some widely used feature selection and feature extraction techniques have analyzed with the purpose of how effectively these techniques can be used to achieve high performance of learning algorithms that ultimately improves predictive accuracy of classifier. An endeavor to analyze dimensionality reduction techniques briefly with the purpose to investigate strengths and weaknesses of some widely used dimensionality reduction methods is presented.

Social networking sites such as Facebook or Twitter attract millions of users, who everyday post an enormous amount of content in the form of tweets, comments and posts. Since social network texts are usually short, learning tasks have to deal with a very high dimensional and sparse feature space, in which most features have low frequencies. As a result, extracting useful knowledge from such noisy data is a challenging task, that converts large-scale short-text learning tasks in social environments into one of the most relevant problems in machine learning and data mining. Feature selection is one of the most known and commonly used techniques for reducing the impact of the high dimensional feature space in text learning. A wide variety of feature selection techniques can be found in the literature applied to traditional, long-texts and document collections. However, short-texts coming from the social Web pose new challenges to this well-studied problem as texts’ shortness offers a limited context to extract enough statistical evidence about words relations (e.g. correlation), and instances usually arrive in continuous streams (e.g. Twitter timeline), so that the number of features and instances is unknown, among other problems. This paper surveys feature selection techniques for dealing with short texts in both offline and online settings. Then, open issues and research opportunities for performing online feature selection over social media data are discussed.

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.

The paper proposes a hybrid approach to reduce dimension in text classification problems, to overcome the issue of Curse of Dimensionality. This hybrid approach is a combination of Feature Selection (FS) and Feature Extraction (FE) methods, considering different aspects of feature relevance, to effectively reduce the dimension in large text datasets. It prevents feature selection biased in favor of a particular FS method. Many FS methods like Term Variance, Document Frequency, Information Gain, Shannons Entropy measure, Mean-Median and Mean Absolute Difference, were implemented and a comparative study was made on their performance when implemented in a hybrid system. The features selected by the individual FS methods are merged using three approaches, namely, Union, Intersection and Modified Union. The sub lists of features further undergo Feature Extraction by PCA, and the reduced feature sub list is clustered with k-means. Finally, the sentiment-score of the individual clusters are calculated using SentiWordNet database which gives the polarity of the data. The experiments were conducted on the benchmark datasets namely Reuters-21,578 and Classic4. The performance evaluation of the system made using the measures like precision, recall, f-score and accuracy shows that the proposed method has improved performance compared to its competitive methods.

Feature selection (FS), which reduces data dimensionality and enriches the performance of any suggested framework, is one of the most crucial parts of machine learning (ML). In real-world applications, computational, large dimensionality and noisy or ambiguous nature, storage complexity, high performance, and other difficulties beset FS work. As a pre-processing phase, dimension reduction is used to improve the accuracy of learning features and reduce training time by removing irrelevant data, noise, and duplicated features. The feature selection (FS) and feature extraction (FE) methods were used to conduct dimensionality reduction (DR). Due to the increasing rate of data creation, FS is a practical method for addressing several serious dimensionality problems, like decreasing duplication, eradicating unneeded data, and enhancing comprehensibility in the outcomes. The efficiency of data processing and storage can be enriched by FE that handles the challenge of selecting the most distinct, instructive, and constrained collection of features. This research focuses on contemporary feature selection and feature extraction algorithms that aid in reducing data dimensionality.KeywordsFeature selectionMachine learningFeature extractionDimensionality reductionData processing

A modified genetic algorithm and weighted principal component analysis based feature selection and extraction strategy in agriculture

In order to solve the problem of face recognition, the method of feature extraction and feature selection is presented in this paper. First using Gabor filters and face image as the convolution Operator to extract the Gabor feature vector of the image and also to uniform sampling; then using the PCA + LDA method to reduce the dimension for high-dimensional Gabor feature vector; Finally, using the nearest neighbor classifier to discriminate and determine the identity of a face image. The result I get is that the sampled Gabor feature in high-dimensional space can be projected onto low-dimensional space though the method of feature selection and compression. The new and original in this paper is that the method of PCA + LDA overcomes the problem of the spread matrix singular in the class and matrix too large which is brought by directly use the LDA.

Today, video data, as a powerful multimedia component, is accompanied by some problems with increasing usage in communication, health, education, and social media in particular. Classification and detection of concepts in video data by automatic methods are some of these challenging problems. In this study, we propose a video classification system, which incorporates deep convolutional neural networks (CNNs) by leveraging feature selection and data fusion techniques to improve the accuracy of the classification. Principal Component Analysis (PCA) as a feature selection method and Discriminant Correlation Analysis (DCA) technique, which incorporates class associations into the correlation analysis of feature sets for data fusion, are applied to the problem at the feature level. Support Vector Machines (SVMs) have been trained with new feature vectors obtained from different deep convolutional neural networks by feature selection and data fusion methods. The proposed method is tested for 38 concepts on TRECVID 2013 SIN video task dataset and the results are evaluated. Our results show that the classification accuracy is improved by 4% with an accuracy of 50.27% when the proposed data fusion and feature selection techniques are used.

In many real applications, the collected data is usually high dimensionality of feature space and no labels are given, which will result in the “curse of dimensionality” problem and bring the challenge for the feature selection. Although tremendous efforts have been made in unsupervised feature selection, the existing methods neglect the robustness of jointly feature selection and clustering. In this paper, we present a robust simultaneous feature selection and clustering method. By introducing a robust loss function, the unsupervised linear discriminant analysis methods with two different strategies (the ratio trace and the trace ratio) are incorporated into this loss function, which can guarantee the selected features with robustness. Then two novel simultaneous unsupervised feature selection and clustering objective functions are presented by the proposed loss function equipped with $\ell_{2,1}$-norm. Two alternative iterative optimization algorithms are developed to solve these two objective functions. The comparative experimental results conducted on eight UCI datasets show that the proposed methods are better than other competing methods.

In the field of fault diagnosis, machine learning is highly valued for its broad applicability and efficiency. Feature extraction and feature selection are key steps in the application of machine learning, and the performance of fault diagnosis methods relies heavily on the effective execution of these two steps. For this reason, this paper aims to enhance the performance of fault diagnosis methods by improving these two aspects. Firstly, to address the non-linearity and non-stationarity of rotating machinery vibration signals under variable operation conditions, this paper proposes an improved rapid refined composite multiscale sample entropy (IR2CMSE) feature extraction method. In addition, this paper decomposes the vibration signals with improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) and extracts the sensitive intrinsic modal functions’ (IMFs') IR2CMSE values (SI-IR2CMSE) as the initial feature vector, which more accurately reveals the intrinsic time-scale characteristics of the vibration signals. Secondly, to address the problem of over-reliance on sample labels in most feature selection methods, this paper proposes a semi-supervised Gaussian mixing model with sparse regularization (SSGMM-SR) feature selection model. The model does not require complete fault labels and can automatically identify important features. Finally, validation with two rotating machinery fault datasets shows that the method proposed in this study exhibits high diagnostic accuracy and stability across multiple classifiers.

To recognize expressions accurately, facial expression systems require robust feature extraction and feature selection methods. In this paper, a normalized mutual information based feature selection technique is proposed for FER systems. The technique is derived from an existing method, that is, the max-relevance and min-redundancy (mRMR) method. We, however, propose to normalize the mutual information used in this method so that the domination of the relevance or of the redundancy can be eliminated. For feature extraction, curvelet transform is used. After the feature extraction and selection the feature space is reduced by employing linear discriminant analysis (LDA). Finally, hidden Markov model (HMM) is used to recognize the expressions. The proposed FER system (CNF-FER) is validated using four publicly available standard datasets. For each dataset, 10-fold cross validation scheme is utilized. CNF-FER outperformed the existing well-known statistical and state-of-the-art methods by achieving a weighted average recognition rate of 99 % across all the datasets.

Feature selection is a very important aspect in the field of machine learning. It entails the search of an optimal subset from a very large data set with high dimensional feature space. Apart from eliminating redundant features and reducing computational cost, a good selection of feature also leads to higher prediction and classification accuracy. In this paper, an efficient feature selection technique is introduced in the task of epileptic seizure detection. The raw data are electroencephalography (EEG) signals. Using discrete wavelet transform, the biomedical signals were decomposed into several sets of wavelet coefficients. To reduce the dimension of these wavelet coefficients, a feature selection method that combines the strength of both filter and wrapper methods is proposed. Principal component analysis (PCA) is used as part of the filter method. As for wrapper method, the evolutionary harmony search (HS) algorithm is employed. This metaheuristic method aims at finding the best discriminating set of features from the original data. The obtained features were then used as input for an automated classifier, namely wavelet neural networks (WNNs). The WNNs model was trained to perform a binary classification task, that is, to determine whether a given EEG signal was normal or epileptic. For comparison purposes, different sets of features were also used as input. Simulation results showed that the WNNs that used the features chosen by the hybrid algorithm achieved the highest overall classification accuracy.

In text classification, taking words in text documents as features creates a very high dimensional feature space. This is known as the high dimensionality problem in text classification. The most common and effective way to solve this problem is to select an ideal subset of features using a feature selection approach. In this paper, a new feature selection approach called Rough Information Gain (RIG) is presented as a solution to the high dimensionality problem. Rough Information Gain extracts hidden and meaningful patterns in text data with the help of Rough Sets and computes a score value based on these patterns. The proposed approach utilizes the selection strategy of the Information Gain Selection (IG) approach when pattern extraction is completely uncertain. To demonstrate the performance of the Rough Information Gain in the experimental studies, the Micro-F1 success metric is used to compare with Information Gain Selection (IG), Chi-Square (CHI2), Gini Coefficient (GI), Discriminative Feature Selector (DFS) approaches. The proposed Rough Information Gain approach outperforms the other methods in terms of performance, according to the results.

Performance comparison of feature selection and extraction methods with random instance selection