An improved k-prototypes clustering algorithm for mixed numeric and categorical data

Ji, Jinchao; Tian, Bin; Zhou, Chunguang; Ma, Chao; Wang, Zhe

doi:10.1016/j.neucom.2013.04.011

Cited by 106 publications

(76 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The WK-Means algorithm with distributed centroids (WK-DC) was proved to work 60 well on data sets with numerical, categorical and mixed data types [18]. However, as we will show here, there is still room for further improvement of this approach.…”

mentioning

confidence: 94%

“…The latter authors showed that their approach works quite well in a fuzzy scenario, in which a given entity may belong to two or more clusters with different degrees of membership. Ji et al [18], inspired by the concept of fuzzy centroids, has recently introduced the concept 55 of distribution centroids which represent the centers of clusters for categorical features in a crisp scenario, rather than in a fuzzy one. Ji et al [18] also incorporated into their algorithm the variable weight estimation procedure, thus addressing the above-mentioned weaknesses (iii) and (iv) of traditional K-Means.…”

mentioning

confidence: 99%

“…To address this important limitation, Ji et al [18] proposed to use distributed centroids in WK-Means (WK-DC), showing that WK-DC outperforms other popular parti-…”

mentioning

confidence: 99%

“…On the other hand, WK-DC was introduced using solely, and rather successfully, β = 8 [18]. Based on this success, we decided to extend WK-DC by addressing the following three major points.…”

mentioning

confidence: 99%

“…Second, the difference of one between the exponent β and the distance exponent in 125 WK-DC [18] does not allow the feature weights to be seen as feature re-scaling factors. If the two exponents were identical, the weights could be used to re-scale the given data set as a part of data pre-processing step.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Applying subclustering and Lp distance in Weighted K-Means with distributed centroids

Amorim

Makarenkov

2016

Neurocomputing

View full text Add to dashboard Cite

We consider the weighted K-Means algorithm with distributed centroids aimed at clustering data sets with numerical, categorical and mixed types of data. Our approach allows given features (i.e., variables) to have different weights at different clusters. Thus, it supports the intuitive idea that features may have different degrees of relevance at different clusters. We use the Minkowski metric in a way that feature weights become feature re-scaling factors for any considered exponent. Moreover, the traditional Silhouette clustering validity index was adapted to deal with both numerical and categorical types of features. Finally, we show that our new method usually outperforms traditional K-Means as well as the recently proposed WK-DC clustering algorithm.

show abstract

mentioning

confidence: 94%

mentioning

confidence: 99%

“…To address this important limitation, Ji et al [18] proposed to use distributed centroids in WK-Means (WK-DC), showing that WK-DC outperforms other popular parti-…”

mentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

Applying subclustering and Lp distance in Weighted K-Means with distributed centroids

Amorim

Makarenkov

2016

Neurocomputing

View full text Add to dashboard Cite

show abstract

Assessing the degradation of grassland ecosystems based on the advanced local net production scaling method—The case of Inner Mongolia, China

2020

View full text Add to dashboard Cite

The gradual degradation of grasslands on a global scale goes hand‐in‐hand with significant challenges for agriculture and animal husbandry development. Numerous relevant policies and projects have been implemented to protect and restore Chinese ecosystems, but it is still unclear to what degree grassland ecosystems can be recovered. In view of this, we constructed an advanced local net production scaling (ALNS) method by replacing the classification method by self‐organizing feature maps (SOFM) and tailoring the ideal state evaluation method in the LNS method. The ALNS method is used to analyze differences within grassland ecosystems, explore the ideal state of grassland ecosystems, and define degradation as the degree to which the actual state deviates from the ideal state, representing the degree to which grassland ecosystems can be recovered. It thereby quantifies and assesses the overall degradation of such systems in Inner Mongolia. Based on the results, more than 98.5% of the total grassland area failed to reach the ideal state, with the highest levels in the northeast with DN (degraded net primary productivity) values exceeding 200 gc/(m2·yr), followed by the midlands with DN values from 50 to 200 gc/(m2·yr) and the southwest with DN values between 0 and 150 gc/(m2·yr). The ALNS method can efficiently assess grassland ecosystem degradation and can be used to indicate the deviation degrees from ideal states, facilitating the development of protection and restoration programs for grassland ecosystems.

show abstract

Distance‐based clustering of mixed data

Velden

D’Enza

Markos

2018

WIREs Computational Stats

View full text Add to dashboard Cite

Cluster analysis comprises of several unsupervised techniques aiming to identify a subgroup (cluster) structure underlying the observations of a data set. The desired cluster allocation is such that it assigns similar observations to the same subgroup. Depending on the field of application and on domain‐specific requirements, different approaches exist that tackle the clustering problem. In distance‐based clustering, a distance metric is used to determine the similarity between data objects. The distance metric can be used to cluster observations by considering the distances between objects directly or by considering distances between objects and cluster centroids (or some other cluster representative points). Most distance metrics, and hence the distance‐based clustering methods, work either with continuous‐only or categorical‐only data. In applications, however, observations are often described by a combination of both continuous and categorical variables. Such data sets can be referred to as mixed or mixed‐type data. In this review, we consider different methods for distance‐based cluster analysis of mixed data. In particular, we distinguish three different streams that range from basic data preprocessing (where all variables are converted to the same scale), to the use of specific distance measures for mixed data, and finally to so‐called joint data reduction (a combination of dimension reduction and clustering) methods specifically designed for mixed data. This article is categorized under: Statistical Learning and Exploratory Methods of the Data Sciences > Clustering and Classification Statistical Learning and Exploratory Methods of the Data Sciences > Exploratory Data Analysis Statistical and Graphical Methods of Data Analysis > Dimension Reduction

show abstract

An improved k-prototypes clustering algorithm for mixed numeric and categorical data

Cited by 106 publications

References 20 publications

Applying subclustering and Lp distance in Weighted K-Means with distributed centroids

Applying subclustering and Lp distance in Weighted K-Means with distributed centroids

Assessing the degradation of grassland ecosystems based on the advanced local net production scaling method—The case of Inner Mongolia, China

Distance‐based clustering of mixed data

Contact Info

Product

Resources

About