Beatriz Sevilla-Villanueva

Using CVI for Understanding Class Topology in Unsupervised Scenarios

Cluster validation in Clustering is an open problem. The most exploited possibility is the validation through cluster validity indexes (CVIs). However, there are many indexes available, and they perform inconsistently scoring different partitions over a given dataset. The aim of the study carried out is the analysis of seventeen CVIs to get a common understanding of its nature, and proposing an efficient strategy for validating a given clustering. A deep understanding of what CVIs are measuring has been achieved by rewriting all of them under a common notation. This exercise revealed that indexes measure different structural properties of the clusters. A Principal Component Analysis (PCA) confirmed this conceptual classification. Our methodology proposes to perform a multivariate joint analysis of the indexes to learn about the cluster topology instead of using them for simple ranking in a competitive way.

Identifying nutritional patterns through integrative multiview clustering

The main goal of this work is to develop a methodology for finding nutritional patterns based on a variety of subject characteristics which can contribute to better understand the interactions between nutrition and health, provided that the complexity of the phenomenon gives poor performance using classical approaches. An innovative methodology based on advanced clustering techniques is proposed in order to find more compact patterns or clusters. The Integrative Multiview Clustering (IMC) combines Multiview Clustering approach with crossing operations over the several partitions obtained. Comparison with other classical clustering techniques is provided to assess the performance of our approach. The Dunn-like cluster validity index proposed by Bezdek & Pal is used for the comparison from a structural point of view, as it is more robust than the original Dunn index. The performance of the IMC method is better than other popular clustering techniques based on the Dunn-like Index. Our findings suggest that the Integrative Multiview Clustering provides more compact and separated clusters. In addition, IMC helps to reduce the high dimensionality of the data based on multiview division of attributes and also, the resulting partition is easier to interpret. Using the Integrative Multiview Clustering approach, a good partition is obtained from a structural point of view. Also, the interpretation of the resulting partition is clearer than the one obtained by classical approache

Post-processing the Class Panel Graphs: towards understandable patterns from data

A profiling methodology is introduced for automatic interpretation of clusters in this work. This methodology contributes to the characterization of the resulting classes from a clustering process. This work aims to find a concordance between the proposed methodology and the experts’ description of these classes. In this work the resulting classes from a clustering of a general population sample based on their diet and physical activity habits are interpreted and compared with the experts’ description of these classes by using the Class Panel Graphs. In this work, we import techniques from the multivariate analysis into the cluster interpretation process.

Evaluation of Adherence to Nutritional Intervention Through Trajectory Analysis

Classical pre–post intervention studies are often analyzed using traditional statistics. Nevertheless, the nutritional interventions have small effects on the metabolism and traditional statistics are not enough to detect these subtle nutrient effects. Generally, this kind of studies assumes that the participants are adhered to the assigned dietary intervention and directly analyzes its effects over the target parameters. Thus, the evaluation of adherence is generally omitted. Although, sometimes, participants do not effectively adhere to the assigned dietary guidelines. For this reason, the trajectory map is proposed as a visual tool where dietary patterns of individuals can be followed during the intervention and can also be related with nutritional prescriptions. The trajectory analysis is also proposed allowing both analysis: 1) adherence to the intervention and 2) intervention effects. The analysis is made by projecting the differences of the target parameters over the resulting trajectories between states of different time-stamps which might be considered either individually or by groups. The proposal has been applied over a real nutritional study showing that some individuals adhere better than others and some individuals of the control group modify their habits during the intervention. In addition, the intervention effects are different depending on the type of individuals, even some subgroups have opposite response to the same intervention.

The role of significance tests in consistent interpretation of nested partitions

Cluster interpretation is an important step for a proper understanding of a set of classes, independently of whether they have been automatically discovered or expert-based. An understanding of classes is crucial for the further use of classes as the basis of a decision-making process.The abundant work on cluster validity found in the literature is mainly focused on the validation of clusters from the structural point of view. However, structural validation does not ensure that the clustering is useful, since meaningfulness is the key to guaranteeing that classes can support further decisions. In previous works, special significance tests taken from the field of multivariate analysis were introduced in an interpretation methodology for automatically assessing relevant variables in particular classes.In this paper, we present the interpretation of nested partitions and the relationships between both interpretations are studied. In particular, the inconsistencies produced in interpretation when a second partition refines the first one with a higher level of granularity are studied, diagnosed, and a modification of the original methodology is provided to guarantee consistency in these cases. The relevant characteristics detected in a parent class must also be inherited in subclasses, or at least in some of them.The proposal is evaluated using a real data set on baseline health conditions and dietary habits of a sample of the general population.

Estimation of Machine Settings for Spinning of Yarns – New Algorithms for Comparing Complex Structures

The textile industry in Europe is facing a new challenge in order to stay competitive into the textile market. They need to be flexible, cost efficient and produce with high quality. The setting of the machinery parameters is therefore an important aspect that combines implicit knowledge of workers and engineers with explicit knowledge. This makes it an ideal domain for CBR. It is used for an automatic parameter setting but the data cannot be reduced to a flat representation, as yarns and fabrics are multicomponent artefacts. Therefore we propose a combination of 4 algorithms to evaluate the similarity of the yarns. The application was successfully applied for spinning and it can be applied in the following steps of the textile processes like weaving.

Intelligent Management of Measurement Units Equivalences in Food Databases

It is currently well-known that diet plays an important role in the promotion of healthy lifestyle and the prevention of chronic diseases. The Diet4You project is conceived to support the creation of an intelligent decision support system that provides personalized menus fitting a nutritional plan and taking into account the characteristics, needs and preferences of the person. The system involves a background food database, recording a collection of foods and prepared dishes with their standard portions as well as their nutritional decomposition in different food families. This DB is used to search the best combination of dishes approaching the total intake of different nutrients specified in the prescribed nutritional plan. The available background databases, specify the quantities of standard portions of several foods based on different measurement units which are not standardized, and it happens that the weight specified by one cup of melon is different from that of one cup of berries, among others. This arises the need of applying variable conversion factors to the dish description, before assessing whereas the total quantities of a certain menu fit well to the prescription. In this paper, a knowledge based approach is presented to the automatically management. An annotated reference food ontology is built on the basis of additional documentation. However the granularity of the information provided is heterogeneous and non exhaustive. The ontology-based missing values imputation is presented to overcome this limitations.

A methodology to discover and understand complex patterns: Interpreted Integrative Multiview Clustering (I2MC)

The main goal of this work is to develop a methodology for finding nutritional patterns from a variety of individual characteristics which can contribute to better understand the interactions between nutrition and health, provided that the complexity of the phenomenon gives poor performance using classical approaches. An innovative methodology based on a combination of advanced clustering techniques and consistent conceptual interpretation of clusters is proposed to find more understandable patterns or clusters. The Interpreted Integrative Multiview Clustering (I2MC) combines the previously proposed Integrative Multiview Clustering (IMC) with a new interpretation methodology NCIMS. IMC uses crossing operations over the several partitions obtained with the different views. Comparison with other classical clustering techniques is provided to assess the performance of this approach. IMC helps to reduce the high dimensionality of the data based on multiview division of variables. Two innovative Cluster Interpretation methodologies are proposed to support the understanding of the clusters. These are automatic methods to detect the significant variables that describe the clusters; also, a mechanism to deal with the consistency between the interpretations inter clusters of a single partition CI-IMS, or between pairs of nested partitions NCIMS. Some formal concepts are specifically introduced to be used in the NCIMS. I2MC is used to validate the interpretability of the participant’s profiles from an intervention nutritional study. The method has advantages to deal with complex datasets including heterogeneous variables corresponding to different topics and is able to provide meaningful partitions.