Abstract Matching is a research and application domain within machine learning and statistics that provides tools to aggregate, combine, or compare data from different sources, tables, or datasets based on common criteria. This work aims to investigate data matching process applied to Polycystic Ovary Syndrome datasets using low-cost information. Here, we highlight the importance of considering different strategies for data fusion based on statistical tests in order to determine which variables should be integrated into the final dataset. Moreover, we are evaluating how different matching strategies affect the performance of the data-learning models for classifying PCOS. We conduct the experimental evaluation using real data to illustrate the effectiveness of the proposed approach. The performance of various classifiers is measured by different metrics and the results are also discussed in comparison with those corresponding to machine learning literature related to PCOS data.

Abstract This paper introduces a partitioning algorithm for interval data using adaptive distances defined by block variance-covariance matrices. The algorithm has an iterative three-step relocation involving the construction of a partition, identification of a suitable prototype and computation of an adaptive distance by each cluster. In this paper, the distance of each cluster changes at each iteration of the algorithm and it is estimated taking into account joint variance-covariance sub-matrices of the upper and lower boundaries of the interval variables. An experimental evaluation with real and synthetic interval data is performed and the results have confirmed the effectiveness of the proposed algorithm. Moreover, insights about Brazilian interval temperature data clustering are extracted from an application of the algorithm regarding interpretation indices of the cluster analysis for interval data.

Abstract Fake news is information that does not represent reality but is commonly shared on the internet as if it were true, mainly because of its dramatic, appealing, and controversial content. Therefore, a relevant issue is to find characteristics that can assist in identifying Fake News, mainly nowadays, where an increasing number of fake news is spread all over the internet every day. This work aims to extract knowledge from Brazilian fake news data based on statistical learning. Initially, an exploratory data analysis is performed for the available variables to extract insights from the differences between fake and true news. Then, the prediction and modelling are carried out. The learning phase aims to build a model and measure the features that best explain the behaviour of misleading texts, which leads to a parsimonious model. Finally, the test phase estimates the fitted model accuracy based on 10‐fold cross‐validation in the Monte Carlo framework. The results show that four variables are significant to explain fake news. Moreover, our model achieved comparable results with state‐of‐the‐art, 0.941 F‐measure, for a single classifier while having the advantage of being a parsimonious model. This work's details and code can be found at https://github.com/limagbz/fake-news-ptBR .

Millions of papers are submitted and published every year, but researchers often do not have much information about the journals that interest them. In this paper, we introduced the first dynamical clustering algorithm for symbolic polygonal data and this was applied to build scientific journals profiles. Dynamic clustering algorithms are a family of iterative two-step relocation algorithms involving the construction of clusters at each iteration and the identification of a suitable representation or prototype (means, axes, probability laws, groups of elements, etc.) for each cluster by locally optimizing an adequacy criterion that measures the fitting between clusters and their corresponding prototypes The application gives a powerful vision to understand the main variables that describe journals. Symbolic polygonal data can represent summarized extensive datasets taking into account variability. In addition, we developed cluster and partition interpretation indices for polygonal data that have the ability to extract insights about clustering results. From these indices, we discovered, e.g., that the number of difficult words in abstract is fundamental to building journal profiles.

Background: Teamwork is central component of any software development organization. Therefore, the assessment of teamwork quality is important for team management in practice. TWQ (Teamwork Quality) is a measure of the quality of intra-team interactions developed specifically for software development teams. Aims: To perform a differentiated replication of previous studies on TWQ to expand the contexts in which TWQ has been applied, to refine the measurement instrument, increasing its reliability and validity, and to identify possibilities for future research on software teams. Method: We performed a cross-sectional survey collecting data from all 18 teams in one single software organization and from all members of each team, totaling 123 participants. Results: First, we refined the measurement instrument, achieving a more reliable and parsimonious instrument. Second, our results confirmed findings from previous studies when individual level data was used, but almost no confirmation was found when data was aggregated at team level. Conclusions: Our results partially supported previous studies but raised questions about the validity of the aggregation of individual data to team level measures of the studied constructs.

This work aimed to appraise a multivariate time series, high-dimensionality data-set, presented as intervals using a Symbolic Data Analysis (SDA) approach. SDA reduces data dimensionality, considering the complexity of the model information through a set-valued (interval or multi-valued). Additionally, Dynamic Linear Models (DLM) are distinguished by modeling univariate or multivariate time series in the presence of non-stationarity, structural changes and irregular patterns. We considered neurophysiological (EEG) data associated with experimental manipulation of verticality perception in humans, using transcranial electrical stimulation. The innovation of the present work is centered on use of a dynamic linear model with SDA methodology, and SDA applications for analyzing EEG data.

Usually, in a fuzzy clustering, the memberships are the same for all the variables (features), i.e., the variables are considered equally important for the definition of the memberships. Fuzzy Kohonen Clustering network (FKCN) is a self-organizing fuzzy neural network that uses fuzzy membership values from the popular Fuzzy c-Means as learning rates. The replacement of the arbitrary learning rate by a fuzzy membership function can produce better clustering results. This paper introduces a new variant of the FKCN algorithm that finds a set of weights and a multivariate fuzzy partition minimizing an objective function. Here, the multivariate memberships allow to take account the intra-class and inter-class dispersion structures of the input data. Experiments with different configurations of synthetic data sets and applications with real data sets demonstrate the usefulness of this fuzzy clustering network model.

Fuzzy c-Means (FCM) and Possibilistic c-Means (PCM) are the most popular algorithms of the fuzzy and possibilistic clustering approaches, respectively. A hybridization of these methods, called Possibilistic Fuzzy c-Means (PFCM), solves noise sensitivity defect of FCM and overcomes the coincident clusters problem of PCM. Although PFCM have shown good performance in cluster detection, it does not consider that different variables can produce different membership and possibility degrees and this can improve the clustering quality as it has been performed with the Multivariate Fuzzy c-Means (MFCM). Here, this work presents a generalized multivariate approach for possibilistic fuzzy c-means clustering. This approach gives a general form for the clustering criterion of the possibilistic fuzzy clustering with membership and possibility degrees different by cluster and variable and a weighted squared Euclidean distance in order to take into account the shape of clusters. Six multivariate clustering models (special cases) can be derivative from this general form and their properties are presented. Experiments with real and synthetic data sets validate the usefulness of the approach introduced in this paper using the special cases.

Fuzzy C-Means (FCM) is the most popular algorithm of the fuzzy clustering approach. Although FCM and its variations have shown good performance in cluster detection, they do not consider that different variables could produce different membership degrees. Motivated by this, the Multi-variate Fuzzy C-Means (MFCM) method was proposed. The MFCM computes membership degrees of both clusters and variables. However, up to now, there is no method that interprets the multivariate membership of the variables for aiding the understanding of clusters. The goal of this paper is, thus, to bridge this gap by proposing a method to do so. In order to illustrate the method proposed, experiments with a synthetic dataset and applications concerning cancer gene expression data are carried out.

Linear regression is a standard statistical method widely used for prediction. It focuses on modeling the mean the target variable without accounting for all the distributional properties of this variable. In contrast, the quantile regression model facilitates the analysis of the full distributional properties, it allows to model different quantities of the target variable. This paper proposes a quantile regression model for interval data. In this model, each interval variable of the input data is represented by its range and center and a smooth function between two vectors composed by interval variables are defined. In order to test the usefulness of the proposed model, a simulation study is undertaken and an application using a scientific production interval data set of institutions from Brazil is performed. The quality of the interval prediction obtained by the proposed model is assessed by mean magnitude of relative error calculated from test data.

The main property of kernel methods is that they can implicitly perform a nonlinear mapping of the input data into a high-dimensional space. This mapping allows to find a simpler structure within space without increasing the number of parameters increasing the clustering quality. Therefore, kernel methods may find better results for data arranged not linearly. Many methods presented in the literature only use point data. However, real problems need more complex representation. In this work, we propose a new kernel-based fuzzy method using feature space metric for interval-valued data. Moreover, a comparative study between input space and feature space is set up in this paper. In order to evaluate the performance of the proposed method, experiments with synthetic and real interval data set were carried out.

Outliers may have many anomalous causes, for example, credit card fraud, cyberintrusion or breakdown of a system. Several research areas and application domains have investigated this problem. The popular fuzzy c-means algorithm is sensitive to noise and outlying data. In contrast, the possibilistic partitioning methods are used to solve these problems and other ones. The goal of this paper is to introduce cluster algorithms for partitioning a set of symbolic interval-type data using the possibilistic approach. In addition, a new way of measuring the membership value, according to each feature, is proposed. Experiments with artificial and real symbolic interval-type data sets are used to evaluate the methods. The results of the proposed methods are better than the traditional soft clustering ones.

Symbolic data analysis deals with complex data types, capable of modeling internal data variability and imprecise data. This paper introduces two Fuzzy Learning Vector Quantization algorithms for interval symbolic data. One algorithm employs an interval Euclidean distance. The second uses a weighted interval Euclidean distance to try and achieve a better performance of classification when the data set is composed of classes with varying sizes, shapes and structures. The algorithms are evaluated for their performances with synthetic and real data sets. This paper aims at contributing to the area of Supervised Learning within Symbolic Data Analysis.

This article presents a new method for shape description suitable to be used as a solution to the retrieval problem in large image collections. The proposed approach, called Multiscale Symbolic Data Descriptor (MSDD) combines multiscale methods with Symbolic Data Analysis. The contour convexities and concavities at different scale levels are represented using a two-dimensional matrix from which we extract Symbolic Data in order to have a compact and efficient representation of the image in terms of computational resources.

This paper proposes a nonparametric multiple regression method for interval data. Regression smoothing investigates the association between an explanatory variable and a response variable. Here, each interval variable of the input data is represented by its range and center and a smooth function between a pair of vector of interval variables is defined. In order to test the suitability of the proposed model, a simulation study is undertaken and an application using thirteen project data of the NASA repository to estimate interval software size is also considered. These real data represent variability and/or uncertainty innate to the project data. The prediction quality is assessed by a mean magnitude of relative errors calculated from test data.

A novel extension of the fuzzy c-means clustering algorithm for interval data type based on an adaptive Euclidean distance is presented. The proposed method furnishes a fuzzy partition and a prototype for each cluster by optimizing a criterion based on an adaptive Euclidean distance that changes at each algorithm iteration. Experiments with real and synthetic data sets show the usefulness of this method.

The recording of interval data has become a common practice in real world applications and nowadays this kind of data is often used to describe objects. In this paper, we introduce a new fuzzy Kohonen clustering network for symbolic interval data (IFKCN). The network combine the idea of fuzzy membership values for learning rates and the algorithm is able to show superiority in processing the ambiguity and the uncertainty present in data sets. Experiments with benchmark interval data sets and an artificial interval data set for evaluating the usefulness of the proposed method were carried out.

This article presents a new method for texture description suitable to be used as a solution to the retrieval problem in large image collections. The proposed approach combines multiscalegray-level co-occurrence matrices (GLCM) with Symbolic Data Analysis. A benchmark data set is used to demonstrate the usefulness of the proposed methodology. The experimental results demonstrate that the proposed method is encouraging with an average successful rate of 100% for Dataset 1 and 97.9% for Dataset 2.

Automatic vehicle classification is an important task in Intelligent Transport System (ITS) because it allows to obtain the traffic parameter called vehicles count by category. In terrestrial public roads variants sources of information, for vehicles counter by category, have been used such as video, magnetic induction coil, sound sensors, temperature sensors and microwave. The use of video has increased support for traffic management due to the advantages of installation cost and a wide range of information it contains. However, proposed methods of images vehicles classification, obtained from videos of roads traffic, have known limitations, such as strong dependence of detection methods, hard image normalization, noise and low accuracy. This paper presents a method for image vehicle classification based on road traffic video, whose objectives are easy normalization and acceptable accuracy. The method consists of three stages: normalization, training and classification. The images were obtained from road traffic video, taken during a summer day. Edge and PCA like features and the Adaptive-KNN like distance were used for classification. The experimental platform is built on Matlab R2009a.

Clustering analysis is an important tool used in several application domains like pattern recognition, computer vision and computational biology to summarize data. The fuzzy c-means method (FCM) is the most popular fuzzy clustering algorithm, however this method is sensitive to noisy data. The possibilistic c-means (PCM) was created as an alternative to solve this problem. The propose in this work is extend the classical PCM to symbolic interval-valued data. Experiments with artificial and real symbolic interval-type data sets are performed and the results show the superiority of PCM in relation to FCM methods to interval-valued data.

To solve the problem of situations with nonlinearly separable clusters, kernel clustering methods have been proposed. Symbolic Data Analysis (SDA) has emerged to deal with variables that can have intervals, histograms, and even functions as values, in order to consider the variability and/or uncertainty innate to the data. In this paper, we present a K-means clustering method based in kernelized squared L2 distance for symbolic interval-type data. Experiments with real and syntectic symbolic interval-type data sets are considered.

Kernel clustering methods have been very important in application of non-supervised machine learning to real problems. Kernel methods possess many advantages other than non-linearity such as modularity, ability to work with heterogeneous descriptions of data, incorporation of prior knowledge etc. In this paper, we present a clustering method based on kernel functions for partitioning a set of interval-valued data. In addition, this method is compared to a fuzzy partitioning approach for interval data introduced previously. Experiments with real and syntectic symbolic interval-valued data sets are presented. The evaluation of the clustering results furnished by the methods is performed regarding the computation of an external cluster validity index and the global error rate of classification.

This paper presents learning vector quantization classifiers with adaptive distances. The classifiers furnish discriminant class regions from the input data set that are represented by prototypes. In order to compare prototypes and patterns, the classifiers use adaptive distances that change at each iteration and are different from one class to another or from one prototype to another. Experiments with real and synthetic data sets demonstrate the usefulness of these classifiers.

A K-medoids clustering algorithm for mixed feature-type symbolic data represented by categorical, interval-valued and histogram-valued is presented in this paper. The algorithm furnishes a partition and a prototype to each class by optimizing an adequacy criterion based on a suitable standardized Euclidean distance. To evaluate the proposed algorithm, several real symbolic data sets are considered and the results furnished by this algorithm are compared with the results furnished by a partitional algorithm for mixed feature-type symbolic data of the literature of symbolic data analysis in terms of the correct Rand index.

In this paper we propose is an extension of kernel k-means clustering algorithm for symbolic interval data with aggregated kernel functions. To evaluate this method, experiments with synthetic interval data set was performed and we have been compared our method with a dynamic clustering algorithm with single adaptive distance. The evaluation is based on an external cluster validity index (corrected Rand index) and the overall error rate of classification (OERC). This experiment showed the usefulness of the proposed method and the results indicate that aggregated kernel clustering algorithm gives markedly better performance on data sets considered.

Kernel k-means algorithms have recently been shown to perform better than conventional k-means algorithms in unsupervised classification. In this paper we present is an extension of kernel k-means clustering algorithm for symbolic interval data. To evaluate this method, experiments with synthetic and real interval data sets were performed and we have been compared our method with a dynamic clustering algorithm with adaptive distance. The evaluation is based on an external cluster validity index (corrected Rand index) and the overall error rate of classification (OERC). These experiments showed the usefulness of the proposed method and the results indicate that kernel clustering algorithm gives markedly better performance on data sets considered.

This article presents a hybrid approach for texture-based image classification using the gray-level co-occurrence matrices (GLCM) and self-organizing map (SOM) methods. The GLCM is a matrix of how often different combinations of pixel brightness values (grey levels) occur in an image. The GLCM matrices extracted from an image database are processed to create the training data set for a SOM neural network. The SOM model organizes and extracts prototypes from processed GLCM matrices. This paper proposes a novel strategy to index match scores by searching through prototypes. A benchmark data set is used to demonstrate the usefulness of the proposed methodology. The evaluation of performance is based on accuracy in the framework of a Monte Carlo experience. This approach is compared with several classifiers in Li et al. The experimental results on the Brodatz texture image database demonstrate that the proposed method is encouraging with an average successful rate of 97%.

This paper introduces a robust prediction method for symbolic interval data based on the simple linear regression methodology. Each example of the data set is described by feature vector, for which each feature is an interval. Two classic robust regression models are fitted, respectively for range and mid-points of the interval values assumed by the variables in the data set. The prediction of the lower and upper bounds of the new intervals is performed from these fits. To validate this model, experiments with a synthetic interval data set and an application with a cardiology interval-valued data set are considered. The fit and prediction qualities are assessed by a pooled root mean square error measure calculated from learning and test data sets, respectively.

This paper discusses nearest-neighborhood linear regression methods in a statistical view of learning and present an application of these models to software project effort estimation. The usefulness of the models is highlighted through experiments with a well-known NASA software project data set. A comparative study with global regression methods such as bagging predictors, support vector regression, radial basis functions neural networks is also introduced.

Ensemble methods like bagging combine the decisions of multiple classifiers in order to obtain more accuracy than a single classifier. This paper studies the use of bagging for a region oriented symbolic classifier. Experiments with two artificial data sets, generated according to bi-variate normal distributions have been performed in order to show the usefulness of bagging for this symbolic classifier. The prediction accuracy (error rate) of the proposed ensemble is calculated through a Monte Carlosimulation method with 100 replications.

In a previous work, we presented an approach for shape-based image retrieval using the curvature scale space (CSS) and self-organizing map (SOM) methods. Here, we examine the robustness of the representation with images under different scales. The shape features of images are represented by CSS images extracted from, for example, a large database and represented by median vectors that constitutes the training data set for a SOM neural network which, in turn, will be used for performing efficient image retrieval. Experimental results using a benchmark database are presented to demonstrate the usefulness of the proposed methodology. The evaluation of performance is based on accuracy and retrieval time assessed in the framework of a Monte Carlo experience.

This paper introduces a new classifier based on the multi-class logistic regression model for interval symbolic data. Each example of the learning set is described by a feature vector, for which each feature value is an interval. Two versions of this classifier are considered. First fits a multi-class logistic regression model conjointly on the lower and upper bounds of the interval values assumed by the variables in the learning set. Second fits a multi-class logistic model on the lower and upper bounds separately. The prediction of the class for new examples is accomplished from the computation of the posterior probabilities of the classes. To show the usefulness of this method, examples with synthetic interval symbolic data sets with overlapping classes are considered. The assessment of the proposed classification method is based on the estimation of the average behaviour of the error rate in the framework of the Monte Carlo method.

In a previous work [1], we presented an approach for shape-based image retrieval using the curvature scale space (CSS) and self-organizing map (SOM) methods. Here, we examine the robustness of the representation under affine transforms. Moreover, the CSS images extracted from a database are processed and described by median vectors that constitutes the training data set for a SOM neural network. This way of description improves the accuracy of image retrieval in comparison with the previous work [1] that used the first principal component of the PCA technique. Experiments with a benchmark database are carried out to demonstrate the usefulness of the proposed methodology.

This work presents a clustering method for mixed feature-type symbolic data. The presented method needs a previous pre-processing step to transform mixed symbolic data into modal symbolic data. The dynamic clustering algorithm with adaptive distances has then as input a set of vectors of modal symbolic data (weight distributions) and furnishes a partition and a prototype to each class by optimizing an adequacy criterion that measures the fitting between the clusters and their representatives based on adaptive squared Euclidean distances. Examples with synthetic symbolic data sets and an application with a real symbolic data sets show the usefulness of this method.

The recording of symbolic interval data has become popular with the recent advances in database technologies. This paper introduces a dynamic clustering method to partitioning symbolic interval data. This method furnishes a partition and a prototype for each cluster by optimizing an adequacy criterion that measures the fitting between the clusters and their representatives. To compare symbolic interval data, the method uses a single adaptive Hausdorff distance that at each iteration changes but is the same for all the clusters. Experiments with real and synthetic symbolic interval data sets showed the usefulness of the proposed method.