Performance Of Classification Algorithms Research Articles

Trademark Infraction Prevention Policies and Challenges

The objective of this paper is identification and analysis of the prevention policies of trademark infraction and challenges to find similarity between trademarks. Additionally, this paper proposed an approach to enhance semantic retrieval system of conceptually similar trademarks using algorithms of machine learning like Naive Bayes (NB), Artificial Neural Network (ANN) and Support Vector Machine (SVM). Similarity of trademarks is calculated using Tversky index, Cosine similarity, Jaccard coefficient etc. The performance of classification algorithms are compared on the parameter like accuracy on a same set of trademarks representing real trademark infraction cases. The proposed approach is the first step to automate the process of finding conceptually similar trademarks.

HRNeuro-fuzzy: Adapting neuro-fuzzy classifier for recurring concept drift of evolving data streams using rough set theory and holoentropy

Data stream classification plays a vital role in data mining techniques which extracts the most important patterns from the real world database. Nowadays, many applications like sensor network, video surveillance and network traffic generate a huge amount of data streams. Due to the ambiguity in input data, imprecise input information and concept drift, some problems arise in classifying the data stream. To resolve these problems, we propose a HRNeuro fuzzy system in this paper based on rough set theory and holoentropy function. At first, the input database is given to the PCA algorithm to reduce the dimension of the data. An adaptive neuro fuzzy classifier is utilized where the designing of membership function and rule base are the two important aspects. Then, neuro-fuzzy system undergoes updating when the change of detection occurs between the data streams. Here, the updating behaviour of membership function and rules are performed using rough set theory and holoentropy function. The experimental results are evaluated for the datasets and the performance is analysed by some metrics and compared with the existing systems such as JIT adaptive K-NN and HRFuzzy system. From the result, it is concluded that our proposed fuzzy classifier attains the higher accuracy of 96% which proves the efficient performance of data stream classification algorithm.

Predicting discontinuation of docetaxel treatment for metastatic castration-resistant prostate cancer (mCRPC) with random forest

Prostate cancer is the most common cancer among men in developed countries. Androgen deprivation therapy (ADT) is the standard treatment for prostate cancer. However, approximately one third of all patients with metastatic disease treated with ADT develop resistance to ADT. This condition is called metastatic castrate-resistant prostate cancer (mCRPC). Patients who do not respond to hormone therapy are often treated with a chemotherapy drug called docetaxel. Sub-challenge 2 of the Prostate Cancer DREAM Challenge aims to improve the prediction of whether a patient with mCRPC would discontinue docetaxel treatment due to adverse effects. Specifically, a dataset containing three distinct clinical studies of patients with mCRPC treated with docetaxel was provided. We applied the k-nearest neighbor method for missing data imputation, the hill climbing algorithm and random forest importance for feature selection, and the random forest algorithm for classification. We also empirically studied the performance of many classification algorithms, including support vector machines and neural networks. Additionally, we found using random forest importance for feature selection provided slightly better results than the more computationally expensive method of hill climbing.

Domain Specific Automated Triaging System for Bug Classification

The objective of this paper is to analyze and identify domain specific priority classification of bug reports. Different classification algorithms namely-Linear Discriminant Analysis (LDA), Naive Bayes (NB) to predict the performance measures are used. The performance of classification algorithms are compared using bug report instances of two open source software (OSS) namely NetBeans and Eclipse is downloaded from Bugzilla bug repository. Principal Component Analysis (PCA) for feature selection, Particle Swam Optimization (PSO) for instance selection to reduce the dimension of the bug reports. The results are compared based on four performance measures i.e. Accuracy, Precision, Recall and Processing Time. Feature selection to Instance selection gives better results than Instance selection to Feature selection is the result influenced from experiment performed. This approach shows that linear Discriminant analysis performs better than Naive Bayes classifier.

Evaluating the importance of different communication types in romantic tie prediction on social media

The purpose of this paper is to evaluate which communication types on social media are most indicative for romantic tie prediction. In contrast to analyzing communication as a composite measure, we take a disaggregated approach by modeling separate measures for commenting, liking and tagging focused on an alter’s status updates, photos, videos, check-ins, locations and links. To ensure that we have the best possible model we benchmark 8 classifiers using different data sampling techniques. The results indicate that we can predict romantic ties with very high accuracy. The top performing classification algorithm is adaboost with an accuracy of up to 97.89 %, an AUC of up to 97.56 %, a G-mean of up to 81.81 %, and a F-measure of up to 81.45 %. The top drivers of romantic ties were related to socio-demographic similarity and the frequency and recency of commenting, liking and tagging on photos, albums, videos and statuses. Previous research has largely focused on aggregate measures whereas this study focuses on disaggregate measures. Therefore, to the best of our knowledge, this study is the first to provide such an extensive analysis of romantic tie prediction on social media.

Comparative Analysis of Classification Algorithms on Endometrial Cancer Data

Objective: To expose the Performance of classification algorithms on endometrial cancer data. The best algorithms are listed based on the result of various test options and ranked based on their accuracies. Methods and Analysis: Classification is one of the data mining techniques used to find a model that describes the data classes or concepts. The class-label of strange instance is predicted with the help of classification. It compares the classification algorithms by measuring accuracies, speed and strength of algorithms using WEKA tool. Accuracies of classification algorithms are calculated by means of four different options. The error rate and time taken to build the model also measured. Findings: The accuracies of sixteen algorithms are measured by training set, test set, tenfold cross validation and percentage split testing options. The average accuracies are calculated, then compared and ranked with highest accuracy first. The best five algorithms are taken for final performance on endometrial cancer dataset. The accuracy of Random Forest algorithm is high, but it took 0.16 sec to build the model, whereas the IBK, Random Tree and KStar algorithms’ performs well with 0sec to build the model. Bagging algorithm takes more time to build the model. In terms of time and accuracy IBK produces better results as compared to other algorithms. Random Forest algorithm is most excellent in provisos of correctly classified occurrence. Novelty/Improvement: With the 315 instances of endometrial cancer data, the time taken to build the model is zero for IBK, KStar and Random Tree algorithms. If the number of instance increases then time also will increase.

Remote sensing of land use/cover changes and its effect on wind erosion potential in southern Iran.

Wind erosion is a complex process influenced by different factors. Most of these factors are stable over time, but land use/cover and land management practices are changing gradually. Therefore, this research investigates the impact of changing land use/cover and land management on wind erosion potential in southern Iran. We used remote sensing data (Landsat ETM+ and Landsat 8 imagery of 2004 and 2013) for land use/cover mapping and employed the Iran Research Institute of Forest and Rangeland (IRIFR) method to estimate changes in wind erosion potential. For an optimal mapping, the performance of different classification algorithms and input layers was tested. The amount of changes in wind erosion and land use/cover were quantified using cross-tabulation between the two years. To discriminate land use/cover related to wind erosion, the best results were obtained by combining the original spectral bands with synthetic bands and using Maximum Likelihood classification algorithm (Kappa Coefficient of 0.8 and 0.9 for Landsat ETM+ and Landsat 8, respectively). The IRIFR modelling results indicate that the wind erosion potential has increased over the last decade. The areas with a very high sediment yield potential have increased, whereas the areas with a low, medium, and high sediment yield potential decreased. The area with a very low sediment yield potential have remained constant. When comparing the change in erosion potential with land use/cover change, it is evident that soil erosion potential has increased mostly in accordance with the increase of the area of agricultural practices. The conversion of rangeland to agricultural land was a major land-use change which lead to more agricultural practices and associated soil loss. Moreover, results indicate an increase in sandification in the study area which is also a clear evidence of increasing in soil erosion.

A Comparative Study on Various Data Mining Algorithms with Special Reference to Crop Yield Prediction

Objectives: To compare different data mining algorithms with the same parameters on the 10fold cross validation test to predict the crop yield. Methods/Analysis: Different data mining classification algorithms like K-nearest Neighbor, K-means, Neural Network, Support Vector Machine, Case-based Reasoning, Decision Tree algorithm, etc. are applied for various application of agriculture domain. A comparative study is done by using J48, Naive Bayes and Simple Cart algorithms to determine which classification algorithm is best fitted for crop prediction. Findings: In this study, this work reveals the superior performance of J48 classification algorithm with accuracy 89.33% for crop prediction than the other two classification algorithms Simple Cart and Naive Bayes. Novelty /Improvement: This study first time demonstrates the application of different data mining classification techniques (as discussed above) in the domain of agriculture for yield prediction.

A multiobjective weighted voting ensemble classifier based on differential evolution algorithm for text sentiment classification

Typically performed by supervised machine learning algorithms, sentiment analysis is highly useful for extracting subjective information from text documents online. Most approaches that use ensemble learning paradigms toward sentiment analysis involve feature engineering in order to enhance the predictive performance. In response, we sought to develop a paradigm of a multiobjective, optimization-based weighted voting scheme to assign appropriate weight values to classifiers and each output class based on the predictive performance of classification algorithms, all to enhance the predictive performance of sentiment classification. The proposed ensemble method is based on static classifier selection involving majority voting error and forward search, as well as a multiobjective differential evolution algorithm. Based on the static classifier selection scheme, our proposed ensemble method incorporates Bayesian logistic regression, naïve Bayes, linear discriminant analysis, logistic regression, and support vector machines as base learners, whose performance in terms of precision and recall values determines weight adjustment. Our experimental analysis of classification tasks, including sentiment analysis, software defect prediction, credit risk modeling, spam filtering, and semantic mapping, suggests that the proposed classification scheme can predict better than conventional ensemble learning methods such as AdaBoost, bagging, random subspace, and majority voting. Of all datasets examined, the laptop dataset showed the best classification accuracy (98.86%).

Performance of Activity Classification Algorithms in Free-Living Older Adults.

The objective of this study is to compare activity type classification rates of machine learning algorithms trained on laboratory versus free-living accelerometer data in older adults. Thirty-five older adults (21 females and 14 males, 70.8 ± 4.9 yr) performed selected activities in the laboratory while wearing three ActiGraph GT3X+ activity monitors (in the dominant hip, wrist, and ankle; ActiGraph, LLC, Pensacola, FL). Monitors were initialized to collect raw acceleration data at a sampling rate of 80 Hz. Fifteen of the participants also wore GT3X+ in free-living settings and were directly observed for 2-3 h. Time- and frequency-domain features from acceleration signals of each monitor were used to train random forest (RF) and support vector machine (SVM) models to classify five activity types: sedentary, standing, household, locomotion, and recreational activities. All algorithms were trained on laboratory data (RFLab and SVMLab) and free-living data (RFFL and SVMFL) using 20-s signal sampling windows. Classification accuracy rates of both types of algorithms were tested on free-living data using a leave-one-out technique. Overall classification accuracy rates for the algorithms developed from laboratory data were between 49% (wrist) and 55% (ankle) for the SVMLab algorithms and 49% (wrist) to 54% (ankle) for the RFLab algorithms. The classification accuracy rates for SVMFL and RFFL algorithms ranged from 58% (wrist) to 69% (ankle) and from 61% (wrist) to 67% (ankle), respectively. Our algorithms developed on free-living accelerometer data were more accurate in classifying the activity type in free-living older adults than those on our algorithms developed on laboratory accelerometer data. Future studies should consider using free-living accelerometer data to train machine learning algorithms in older adults.

A Survey on Improving Classification Performance Using Data Pre processing And Machine Learning Methods on NSL-KDD Data

: This paper gives an indication of our study in building rare class prediction models for identifying known intrusions and their variations and anomaly detection schemes for detecting novel attacks whose nature is unknown. Data mining and machine learning have been subjected to general explore in intrusion detection with emphasis on improving the accuracy of detection classifier. The quality of the feature selection methods is one of the important factors that affect the effectiveness of Intrusion Detection scheme (IDS). This paper evaluates the performance of data mining classification algorithms namely C4.5, J48, Nave Bayes, NB-Tree and Random Forest using NSL KDD dataset and focuses on Correlation Feature Selection (CFS) assess. The results demonstrates that NB-Tree and Random Forest outperforms other two algorithms in terms of predictive accuracy and detection rate

Identification of DNA binding proteins using evolutionary profiles position specific scoring matrix

DNA-binding plays a crucial role in different genomics processes including identification of specific nucleotides, regulation of transcription and regulation of gene expression. Various conventional methods have been used for identification of DNA-binding proteins. However, due to large explosion of protein sequences in databases, it is intricate or sometimes impossible to identify DNA-binding proteins. Therefore, it is intensively desired to establish an automated model for identification of DNA binding proteins. In this model, numerical attributes are extracted through Dipeptide composition, Split Amino Acid Composition, and position specific scoring matrix (PSSM). In order to overcome the issue of biasness and reduce true error, oversampling technique SMOTE was applied to balance the datasets. Several classification learners including K-nearest neighbor, Probability Neural Network, Support vector machine (SVM) and Random forest are utilized. Two benchmark datasets and jackknife test are applied to assess the performance of classification algorithms. Among various classification algorithms, SVM achieved the highest success rates in conjunction with PSSM feature space, which are 92.3% accuracy on dataset1 and 88.5% on dataset2. The empirical results revealed that our proposed model obtained the highest results so far in the literatures. It is anticipated that our proposed model might be useful and provides a substance for research and academia community.

An embedded imputation method via Attribute-based Decision Graphs

The performance of classification algorithms is highly dependent on the quality of training data. Missing attribute values are quite common in many real world applications, thus, in such cases, a complementary method to improve the quality of the data and, consequently, promote enhancements of the classifier performance, is necessary. To deal with this problem, two strategies are commonly employed in practice, 1) multiple imputation, which often maintains the statistical properties of the original data and, usually, has good performance, at the expense of high computational costs; 2) single imputation, which, in general, provides a suitable solution for data sets with a few missing attribute values, but hardly achieve good results when the number of missing values is high. This paper proposes a new single imputation method which uses Attribute-based Decision Graphs (AbDG) to estimate the missing values. AbDGs are a new type of data graphs which embed the information contained in the training set into a graph structure, built over pre-defined intervals of values from different attributes. As a consequence, similar data instances induce similar subgraphs when projected onto the AbDG, resulting in distinct patterns of connections. The main contribution of the paper is the proposal of a well-defined procedure to perform imputation, by partially matching instances with missing values against the AbDG. The proposed imputation method can effectively deal with data sets having high rates of missing attribute values while presenting low computational cost; a significant result towards the development of robust expert and intelligent systems. The obtained results show evidences that the proposed method is sound and promote qualitative imputation for classification purposes.

Analytical Study of Selected Classification Algorithms for Clinical Dataset

The objective of this paper is to analyze and identify the best classification solution for clinical decision making. Several classification algorithms Like Discriminant Analysis (LDA), Support Vector Machine (SVM), Artificial Neural Network (ANN), Naive Bayes (NB), and Decision Trees are compared to find the optimum diagnostic accuracy. The performance of classification algorithms are compared using benchmark dataset, breast cancer. The effects of normalization using z-score and min-max approaches are also investigated. The results are compared based on different performance parameters like accuracy, sensitivity, specificity and root node error value. Accuracy has been improved for all classifications methods after normalizing the data set. Z-score normalization performs better for all the measures when compared to min-max normalization. The proposed approach shows higher accuracy rate for Naive Bayes algorithm when compared with the other algorithms. Keywords: Accuracy, Classification, Min-Max Normalization, Sensitivity, Specificity, Z-Score Normalization

An improved ant algorithm with LDA-based representation for text document clustering

Document clustering can be applied in document organisation and browsing, document summarisation and classification. The identification of an appropriate representation for textual documents is extremely important for the performance of clustering or classification algorithms. Textual documents suffer from the high dimensionality and irrelevancy of text features. Besides, conventional clustering algorithms suffer from several shortcomings, such as slow convergence and sensitivity to the initial value. To tackle the problems of conventional clustering algorithms, metaheuristic algorithms are frequently applied to clustering. In this paper, an improved ant clustering algorithm is presented, where two novel heuristic methods are proposed to enhance the clustering quality of ant-based clustering. In addition, the latent Dirichlet allocation (LDA) is used to represent textual documents in a compact and efficient way. The clustering quality of the proposed ant clustering algorithm is compared to the conventional clustering algorithms using 25 text benchmarks in terms of F-measure values. The experimental results indicate that the proposed clustering scheme outperforms the compared conventional and metaheuristic clustering methods for textual documents.

Neighborhood rough set based ECG signal classification for diagnosis of cardiac diseases

Cardiac diseases are one of the foremost reasons of mortality. Hence, the early detection of cardiac diseases based on electrocardiogram (ECG) is important for delivering appropriate and timely treatment to the heart patients and it is increasing the heart patient’s survival. Recent trends in clinical decision making systems appeal automation in ECG signal processing and beat classification. Automatic beat classification is a significant method to support clinical specialists to categorize arrhythmia signals in ECG recording. The main objective of this paper is to construct novel automatic classification system for analysis of ECG signal and decision making purposes. The proposed method involves three main parts: De-noising, feature extraction and classification. Initially, discrete wavelet transform (DWT) is applied before classification for signal De-noising and feature extraction. In this work, neighborhood rough set is applied to classify the ECG signals into normal and four abnormal heart beats. The presence of neighborhood rough set classification algorithm (NRSC) produces very exciting recognition and classification abilities through a wide range of biomedical signal processing. The experimental analysis of the proposed NRSC algorithm is compared with the multi-layered perceptron, decision table, Naive Bayes and J48 classification algorithms. Here, the performance of classification algorithms has been evaluated in terms of sensitivity, specificity, Positive predictive value, negative predictive value, false predictive value, Matthews’s correlation coefficients, F-measure, Folke–Mallows Index and Kulcznski Index. The acquired results showed that the proposed algorithm attained 99.32 % of the classification accuracy using NRSC and DWT. Results indicated that the performance of this proposed NRSC classification method was remarkably superior to that of other classification techniques.

Prediction of Protein Submitochondrial Locations by Incorporating Dipeptide Composition into Chou’s General Pseudo Amino Acid Composition

Mitochondrion is the key organelle of eukaryotic cell, which provides energy for cellular activities. Submitochondrial locations of proteins play crucial role in understanding different biological processes such as energy metabolism, program cell death, and ionic homeostasis. Prediction of submitochondrial locations through conventional methods are expensive and time consuming because of the large number of protein sequences generated in the last few decades. Therefore, it is intensively desired to establish an automated model for identification of submitochondrial locations of proteins. In this regard, the current study is initiated to develop a fast, reliable, and accurate computational model. Various feature extraction methods such as dipeptide composition (DPC), Split Amino Acid Composition, and Composition and Translation were utilized. In order to overcome the issue of biasness, oversampling technique SMOTE was applied to balance the datasets. Several classification learners including K-Nearest Neighbor, Probabilistic Neural Network, and support vector machine (SVM) are used. Jackknife test is applied to assess the performance of classification algorithms using two benchmark datasets. Among various classification algorithms, SVM achieved the highest success rates in conjunction with the condensed feature space of DPC, which are 95.20% accuracy on dataset SML3-317 and 95.11% on dataset SML3-983. The empirical results revealed that our proposed model obtained the highest results so far in the literatures. It is anticipated that our proposed model might be useful for future studies.

Improving Classification Performance with PCA-CFS and PCA-SA Feature Selector

Data mining techniques are increasingly being used in Health care analytics to understand the probability of predicted outcomes. High dimensional data the identification of most relevant features that will best determine the target is a challenging task. Feature selection is aims to remove non-informative features from data mining models. This work is focused on improving the detection performance of classification algorithms by the use of Principal Component Analysis (PCA) based feature extraction and Correlation-based Feature Selection (CFS) and Shapley Values Analysis (SA) based feature ranking approach. The performance of projected hybrid method was evaluated through experiments on six medical datasets with various classifiers. The examined result projected the proposed hybrid feature selectors using the PCA-CFS and PCA-SA improves the detection performance of individual classification algorithms over the other methods in the literature.

Using data Mining to Predict Instructor Performance

During these decades, data mining has become one of the effective tools for data analysis and knowledge management system, so that there are many areas which adapted data mining approach to solve their problems. Using data mining in education to enhance the education system is still relatively new. This paper focuses on predicting the instructor performance and investigates the factors that affect students’ achievements to improve the education system quality. Turkey Student Evaluation records dataset is considered and run on different data classifier such as J48 Decision Tree, Multilayer Perception, Naïve Bayes, and Sequential Minimal Optimization. Comparison of all the four classifiers is conducted to predict the accuracy and to find the best performing classification algorithm among all. The conclusions of this study are very promising and provide another point of view to evaluate student performance. It also highlights the importance of employing data mining tools in the field of education. The results show that using the attribute evaluation method on the dataset increases the prediction performance accuracy.

Leveraging Latent Sentiment Constraint in Probabilistic Matrix Factorization for Cross-domain Sentiment Classification

Sentiment analysis is concerned with classifying a subjective text into positive or negative according to the opinion expressed in it. The performance of traditional sentiment classification algorithms rely heavily on manually labeled training data. However, not every domain has the labeled data because the labeling work is time-consuming and expensive. In this paper, we propose a latent sentiment factorization (LSF) algorithm based on probabilistic matrix factorization technique for cross-domain sentiment classification. LSF works in the setting where there are only labeled data in the source domain and unlabeled data in the target domain. It bridges the gap between domains by exploiting the sentiment correlations between domain-shared and domain-specific words in a two-dimensional sentiment space. Experimental results demonstrate the superiority of our method over the state-of-the-art approaches.