Informative variable identifier: Expanding interpretability in feature selection

Muñoz-Romero, Sergio; Gorostiaga, Arantza; Soguero-Ruíz, Cristina; Mora-Jiménez, Inmaculada; Rojo-Álvarez, José Luis

doi:10.1016/j.patcog.2019.107077

Cited by 23 publications

(12 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the current study, we used it to establish the analysis of average groups of clients in terms of nationality, as well as to scrutinize the specific features of repeater and first-timer clients, which is a key point in hospitality. Nevertheless, one should keep in mind that the use of different approaches to identify relevant features can often provide us with different results, even with the same data input [53,54], as they tend to rely on different optimization criteria. This issue represents a challenge for future works aiming to give stability to the proposal of sets of features in practical applications.…”

Section: Discussionmentioning

confidence: 99%

“…The idea of combining MCA with bootstrap resampling and Parzen windows was first proposed by Corral-De-Witt et al [42], in the context of emergency analysis from alert recordings in 911 units, and that methodology was extended here by stabilizing the resampling method and leveraging kernel methods and domain descriptions in order to give improved confidence volumes. From a Data Science point of view, two highly recommendable directions to cover in hospitality and CRM analysis are the inclusion of machine learning methods providing knowledge discovery in labeled data, such as the Information Variable Identifier method [53] or the use of nonlinear mappings to embeddings, either to higher-dimensional feature spaces (in terms of kernel methods) or to lower-dimensional embedding spaces (in the context of autoencoders and deep learning) [24,62].…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

A Big Data Approach to Customer Relationship Management Strategy in Hospitality Using Multiple Correspondence Domain Description

et al. 2020

Self Cite

View full text Add to dashboard Cite

COVID-19 has hit the hotel sector in a hitherto unknown way. This situation is producing a fundamental change in client behavior that makes crucial an adequate knowledge of their profile to overcome an uncertain environment. Customer Relationship Management (CRM) can provide key strategies in hospitality industry by generating a great amount of valuable information about clients, whereas Big Data tools are providing with unprecedented facilities to conduct massive analysis and to focus the client-to-business relationship. However, few instruments have been proposed to handle categorical features, which are the most usual in CRMs, aiming to adapt the statistical robustness with the best interpretability for the managers. Therefore, our aim was to identify the profiles of clients from an international hotel chain using the overall data in its CRM system. An analysis method was created involving three elements: First, Multiple Correspondence Analysis provides us with a statistical description of the interactions among categories and features. Second, bootstrap resampling techniques give us information about the statistical variability of the feature maps. Third, kernel methods provide easy-to-visualize domain descriptions based on confidence areas in the maps. The proposed methodology can provide an operative and statistically principled way to scrutinize the CRM profiles in hospitality.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

A Big Data Approach to Customer Relationship Management Strategy in Hospitality Using Multiple Correspondence Domain Description

et al. 2020

Self Cite

View full text Add to dashboard Cite

show abstract

“…Post hoc approaches include image perturbation methods applied on the network by masking, substituting features with zero or random counterfactual instances, occlusion, conditional sampling, etc. Such approaches aim at revealing impactful regions in the image that affect the classification result 5 , 6 . Other post hoc methodologies handle the interpretation problem by constructing simple proxy models, with similar behavior to the original model and implement the perturbation notion at a feature-level 7 , 8 .…”

Section: Introductionmentioning

confidence: 99%

E pluribus unum interpretable convolutional neural networks

Dimas¹,

Cholopoulou²,

Iakovidis³

2023

Sci Rep

View full text Add to dashboard Cite

The adoption of convolutional neural network (CNN) models in high-stake domains is hindered by their inability to meet society’s demand for transparency in decision-making. So far, a growing number of methodologies have emerged for developing CNN models that are interpretable by design. However, such models are not capable of providing interpretations in accordance with human perception, while maintaining competent performance. In this paper, we tackle these challenges with a novel, general framework for instantiating inherently interpretable CNN models, named E pluribus unum interpretable CNN (EPU-CNN). An EPU-CNN model consists of CNN sub-networks, each of which receives a different representation of an input image expressing a perceptual feature, such as color or texture. The output of an EPU-CNN model consists of the classification prediction and its interpretation, in terms of relative contributions of perceptual features in different regions of the input image. EPU-CNN models have been extensively evaluated on various publicly available datasets, as well as a contributed benchmark dataset. Medical datasets are used to demonstrate the applicability of EPU-CNN for risk-sensitive decisions in medicine. The experimental results indicate that EPU-CNN models can achieve a comparable or better classification performance than other CNN architectures while providing humanly perceivable interpretations.

show abstract

“…According to the level of the feature index, we can divide the discussion of feature selection into three cases [9], [10]: fixed dimension, divergent dimension, and high dimension. The dimensionality directly affects the selection of feature selection methods.…”

Section: Introduction a Related Workmentioning

confidence: 99%

LSTM-DGMDH: High-Dimensional Index Tracking Based on LSTM and Adaptive Deep Evolutionary GMDH Neural Network

Tong,

Liu,

Liu

et al. 2023

IEEE Access

View full text Add to dashboard Cite

Stock index is an indicator that describes the changes in the total price level of the stock market, and it is susceptible to many dynamic factors, with such characteristics as high dimension, uncertainty, non-linearity, time delay, complexity, etc., resulting in abnormal and missing values in stock index data, which will lead to instability or unreliability of the stock index tracking model. In order to solve these problems, we take the historical stock index as the input, model the internal dynamic changes of features, and learn the change rule. Firstly, we introduce an attention mechanism, that is, to assign different weights to the implicit state of the long short term memory network (LSTM) through mapping weights and learning parameters. We further propose a stock index data preprocessing model of the LSTM based on the attention mechanism. Secondly, the group method of data handling type neural networks (GMDH-NN) is a selforganizing data mining technology, which is especially suitable for modeling complex systems. So we choose a discrete form of Kolmogorov-Gabor (K − G) polynomial of the first-order as the reference function of GMDH-NN to establish the general relationship between input and output variables. We further present a deep evolutionary GMDH polynomial neural network (DGMDH) to perform stock index tracking. Moreover, for a high-dimensional stock index dataset, the traditional external criterion can no longer meet the needs of reality, so we propose a tracking error external criterion (TEEC) for stock indices, which is based on the difference between allocation yield and target yield. The TEEC provides better information for selecting the optimal complex DGMDH model. Our experiments clearly show the effectiveness of our methodology.

show abstract

Informative variable identifier: Expanding interpretability in feature selection

Cited by 23 publications

References 19 publications

A Big Data Approach to Customer Relationship Management Strategy in Hospitality Using Multiple Correspondence Domain Description

A Big Data Approach to Customer Relationship Management Strategy in Hospitality Using Multiple Correspondence Domain Description

E pluribus unum interpretable convolutional neural networks

LSTM-DGMDH: High-Dimensional Index Tracking Based on LSTM and Adaptive Deep Evolutionary GMDH Neural Network

Contact Info

Product

Resources

About