Two-step Clustering Approach Research Articles

DEM-based delineation for improving geostatistical interpolation of rainfall in mountainous region of Central Himalayas, India

In mountainous region with heterogeneous topography, the geostatistical modeling of the rainfall using global data set may not confirm to the intrinsic hypothesis of stationarity. This study was focused on improving the precision of the interpolated rainfall maps by spatial stratification in complex terrain. Predictions of the normal annual rainfall data were carried out by ordinary kriging, universal kriging, and co-kriging, using 80-point observations in the Indian Himalayas extending over an area of 53,484 km2. A two-step spatial clustering approach is proposed. In the first step, the study area was delineated into two regions namely lowland and upland based on the elevation derived from the digital elevation model. The delineation was based on the natural break classification method. In the next step, the rainfall data was clustered into two groups based on its spatial location in lowland or upland. The terrain ruggedness index (TRI) was incorporated as a co-variable in co-kriging interpolation algorithm. The precision of the kriged and co-kriged maps was assessed by two accuracy measures, root mean square error and Chatfield’s percent better. It was observed that the stratification of rainfall data resulted in 5–20 % of increase in the performance efficiency of interpolation methods. Co-kriging outperformed the kriging models at annual and seasonal scale. The result illustrates that the stratification of the study area improves the stationarity characteristic of the point data, thus enhancing the precision of the interpolated rainfall maps derived using geostatistical methods.

Assessing Differences between Early and Later Adopters of Accountable Care Organizations Using Taxonomic Analysis.

To compare early and later adopters of the accountable care organization (ACO) model, using the taxonomy of larger, integrated system; smaller, physician-led; and hybrid ACOs. The National Survey of ACOs, Waves 1 and 2. Cluster analysis using the two-step clustering approach, validated using discriminant analysis. Wave 2 data analyzed separately to assess differences from Wave 1 and then data pooled across waves. Compared to early ACOs, later adopter ACOs included a greater breadth of provider group types and a greater proportion self-reported as integrated delivery systems. When data from the two time periods were combined, a three-cluster solution similar to the original cluster solution emerged. Of the 251 ACOs, 31.1 percent were larger, integrated system ACOs; 45.0 percent were smaller physician-led ACOs; and 23.9 percent were hybrid ACOs-compared to 40.1 percent, 34.0 percent, and 25.9 percent from Wave 1 clusters, respectively. While there are some differences between ACOs formed prior to August 2012 and those formed in the following year, the three-cluster taxonomy appears to best describe the types of ACOs in existence as of July 2013. The updated taxonomy can be used by researchers, policy makers, and health care organizations to support evaluation and continued development of ACOs.

Social-spatial structure of Beijing: a spatial-temporal analysis

In the past couple of post-reform decades of rapid urban growth of Chinese cities, Beijing experienced dramatic spatial and socio-economic transformation. This study aims to examine the evolution of Beijing’s social-spatial structure in the transforming process over the past 20 years. We first collect and process Beijing’s socio-demographic variables at the level of sub-district in three years between 1990 and 2009. Then a two-step spatial clustering approach is taken for spatial data mining of the spatial structures of social areas in the three time periods. Results are visualised in GIS. Finally, the interpretation of the analysis result is provided, followed by a discussion of policy implications for the development of a socially sustainable municipality.

Modelling the ecosystem indicators of British Columbia using Earth observation data and terrain indices

Remotely sensed data plays a critical role by acquiring data on ecological conditions over broad spatial scales, providing important information for mapping landscape-scale ecosystem characteristics. The goal of our research is to employ a robust clustering algorithm to provide a transparent method of integrating remotely sensed datasets into homogeneous ecosystem units for conservation planning and monitoring ecosystem condition and change. Using a suite of ecosystem characteristics derived from digital elevation and remotely sensed data at 1km spatial resolution, we classify the 94 million ha within the province of British Columbia (BC), Canada, into 16 terrestrial ecosystem regions (and a water category) using a two-step clustering approach. Initially, 10 metrics representing the physical environment (elevation and soil wetness potential), available energy (solar insolation and snow melt) and vegetation production (fraction of photosynthetically active radiation) were considered for ecosystem classification, which were reduced to six after analyzing variable inter-correlations. The results provide ecologically unique terrestrial regions: ten of which describe the Northern Boreal, Coastal Mountains and Southern Interior Mountains, and six the coastal lowlands, Georgia Depression, interior, Boreal Plains and Taiga Plains. Analyzing the spatial interaction between the cluster categories revealed that highly dispersed ecosystem types occur most often in the intermediate elevation zone, moderate dispersion at the highest elevations, and homogeneity in the lowland areas where elevation remains relatively constant. When overlaid with BC's standard biogeoclimatic ecosystem classification zones the newly developed regions represent similar ecosystem ranges in the coastal, Taiga and Boreal Plains. However, overall our delineation exhibits a greater level of diversity in the alpine environment, and greater homogeneity in the central and southern interior. The quantitative regionalization approach we present offers a broad-scale assessment of British Columbia's ecosystem diversity that can be used as a supplement to traditional in situ biodiversity assessments to provide detail in under-sampled regions of BC or areas experiencing landscape change.

Use of Clustering Methods to Understand More about the Case

During the past seventy years, the field of cluster analysis has emerged, accompanied by a plethora of methods, algorithms, concepts, and terminology that are used in cluster-related research. We refer to cluster analysis (CA) as a general approach composed of several multivariate methods for delineating natural groups or clusters in data sets. In this paper, we describe the ability of CA to provide rich information about the individual case and highlight potential underlying social processes. First, we discuss the theory behind CA as well as differentiate between more and less familiar clustering approaches. Second, we illustrate the value of less familiar clustering techniques by comparing the results of a four wave growth mixture model of family variables versus clustering the same data with a more familiar two-step approach. The growth mixture modelling approach suggested a one-class cluster solution where all families shared similar growth trajectories in parenting practices and family relationship characteristics. However the two-step clustering approach suggested a four-class solution. Finally, we describe ways that CA allows researchers to model processes whose outcomes are the results of a combination of multiple factors and additional benefits of less familiar clustering methods.

Drawing the lines: resolving fishery management units with simple fisheries data

The task of assessing marine resources should begin with defining management units. Often this step is overlooked or defined at temporal scales irrelevant to management needs. Additionally, traditional methods to define stock structure can be data intensive and (or) cost prohibitive and thus not available for emerging or data-limited fisheries. We present an approach that uses commonly available fisheries data (catch and effort) to delineate management units for dynamically independent populations. Spatially explicit standardized indices of abundance are grouped using a two-step partitioning cluster analysis that includes abundance index uncertainty. This “management unit estimator” is tested via simulation and found generally to recover the true number of management units across data of different temporal length, sample size, and quality. Management units are then determined for four species with varying ecologies, fishery histories, and data issues that exemplify the challenges of applying this method to messy data sets. Defining management units via relative abundance incorporates changes in population connectivity in relation to current removals and environmental conditions and creates consistency of index use within assessments. The two-step clustering approach is simple and widely applicable to situations wherein the clustering metric contains uncertainty.

데이터마이닝에 의한 고객세분화 개발

고객세분화는 기업이 관계하고 있는 고객을 이해하고 그 이해를 바탕으로 효과적인 고객관리를 수행하기 위해 필수적인 요소인데 데이터마이닝이 기업의 정보관리영역에 적극적으로 활용되면서 보다 과학적이고 최적화된 형태로 개발되고 있다. 본 연구에서는 신용카드고객 의 카드사용행태에 근거하여 각 고객을 서로 유사한 사용행태를 보이는 고객군으로 세분화하는 과정을 소개하였다. 고객이 실제로 신용카드를 사용하면서 발생시킨 거래정보에만 의존하여 고객세분화를 개발하였으며 이는 마케팅의 관점에서 상당히 의미있는 내용이 될 수 있다. 고객세분화의 개발을 위하여 데이터마이닝기법인 k-평균 군집방법과 최장연결법에 의한 계보적 군집방법을 단계적으로 활용하는 이단계 군집방법을 이용하였다. To Know customers is very important for the company to survive in its cut-throat competition among coimpetitors. Companies need to manage the relationship with each ana every customer, ant make each of customers as profitable as possible. CRM (Customer relationship management) has emerged as a key solution for managing the profitable relationship. In order to achieve successful CRM customer segmentation is a essential component. Clustering as a data mining technique is very useful to build data-driven segmentation. This paper is concerned with building proper customer segmentation with introducing a credit card company case. Customer segmentation was built based only on transaction data which cattle from customer's activities. Two-step clustering approach which consists of k-means clustering and agglomerative clustering was applied for building a customer segmentation.