K-means Cluster Analysis Method Research Articles

K-means Based Cluster Analysis of Residential Smart Meter Measurements

A clustering module based on the k-means cluster analysis method was developed. Smart meter based residential load profiles were used to validate the clustering module. Several case studies were implemented using daily and segmented load profiles of individual and aggregated smart meters. Simulation results defined in terms of the relationship between the clustering ratio and the segmentation time window reveal that the minimum clustering ratio is obtained for the shortest time window of segmentation. Results also show that a small number of clusters is recommended for highly correlated load profiles.

Risk assessment of water pollution sources based on an integrated k-means clustering and set pair analysis method in the region of Shiyan, China

Source water areas are facing many potential water pollution risks. Risk assessment is an effective method to evaluate such risks. In this paper an integrated model based on k-means clustering analysis and set pair analysis was established aiming at evaluating the risks associated with water pollution in source water areas, in which the weights of indicators were determined through the entropy weight method. Then the proposed model was applied to assess water pollution risks in the region of Shiyan in which China's key source water area Danjiangkou Reservoir for the water source of the middle route of South-to-North Water Diversion Project is located. The results showed that eleven sources with relative high risk value were identified. At the regional scale, Shiyan City and Danjiangkou City would have a high risk value in term of the industrial discharge. Comparatively, Danjiangkou City and Yunxian County would have a high risk value in terms of agricultural pollution. Overall, the risk values of north regions close to the main stream and reservoir of the region of Shiyan were higher than that in the south. The results of risk level indicated that five sources were in lower risk level (i.e., level II), two in moderate risk level (i.e., level III), one in higher risk level (i.e., level IV) and three in highest risk level (i.e., level V). Also risks of industrial discharge are higher than that of the agricultural sector. It is thus essential to manage the pillar industry of the region of Shiyan and certain agricultural companies in the vicinity of the reservoir to reduce water pollution risks of source water areas.

Cluster analysis and its application to healthcare claims data: a study of end-stage renal disease patients who initiated hemodialysis.

BackgroundCluster analysis (CA) is a frequently used applied statistical technique that helps to reveal hidden structures and “clusters” found in large data sets. However, this method has not been widely used in large healthcare claims databases where the distribution of expenditure data is commonly severely skewed. The purpose of this study was to identify cost change patterns of patients with end-stage renal disease (ESRD) who initiated hemodialysis (HD) by applying different clustering methods.MethodsA retrospective, cross-sectional, observational study was conducted using the Truven Health MarketScan® Research Databases. Patients aged ≥18 years with ≥2 ESRD diagnoses who initiated HD between 2008 and 2010 were included. The K-means CA method and hierarchical CA with various linkage methods were applied to all-cause costs within baseline (12-months pre-HD) and follow-up periods (12-months post-HD) to identify clusters. Demographic, clinical, and cost information was extracted from both periods, and then examined by cluster.ResultsA total of 18,380 patients were identified. Meaningful all-cause cost clusters were generated using K-means CA and hierarchical CA with either flexible beta or Ward’s methods. Based on cluster sample sizes and change of cost patterns, the K-means CA method and 4 clusters were selected: Cluster 1: Average to High (n = 113); Cluster 2: Very High to High (n = 89); Cluster 3: Average to Average (n = 16,624); or Cluster 4: Increasing Costs, High at Both Points (n = 1554). Median cost changes in the 12-month pre-HD and post-HD periods increased from $185,070 to $884,605 for Cluster 1 (Average to High), decreased from $910,930 to $157,997 for Cluster 2 (Very High to High), were relatively stable and remained low from $15,168 to $13,026 for Cluster 3 (Average to Average), and increased from $57,909 to $193,140 for Cluster 4 (Increasing Costs, High at Both Points). Relatively stable costs after starting HD were associated with more stable scores on comorbidity index scores from the pre-and post-HD periods, while increasing costs were associated with more sharply increasing comorbidity scores.ConclusionsThe K-means CA method appeared to be the most appropriate in healthcare claims data with highly skewed cost information when taking into account both change of cost patterns and sample size in the smallest cluster.

住宅商品房市场细分—基于A市居民购房意向调研

为了更好地了解消费者的购房需求，细分住宅商品房市场，为房地产商制定精准的市场营销策略提供借鉴，本文将传统的统计学方法和机器学习的方法相结合定量分析了住宅商品房市场的需求状况。首先，文章结合以往研究经验选取房地产市场细分指标，通过问卷调查获取A市居民的购房意向与需求信息。然后，采用K-means聚类分析和决策树分类方法将消费者分为四类。最后，根据分类结果，运用频数分析、百分比分析、交叉列联分析方法得出了每一个类别的市场特征和需求倾向，为房地产开发商精准营销提供借鉴。同时，本文还得出了四类市场群体在住宅户型、购房关注点等方面的选择共性，为房地产开发商初期项目建设也有一定的指导意义。 In order to better understand the housing demand of consumer, divide the residential real estate market, and provide reference for precise marketing strategies for real estate developers, this paper, combining the traditional statistical methods and machine learning method, analyzes the residential housing market demand in quantitative way. First of all, this paper bases previous re-search experience to select indicators for the real estate market segmentation, through the ques-tionnaire survey to obtain A city’s information about residential purchase intention and demand. Then, through using the K-means cluster analysis and decision tree classification method, it comes to the conclusion that consumers can be divided into four groups. Finally, according to the classi-fication results, and using frequency, percentage, cross contingency analysis methods, it shows demand characteristics and tendencies of each category. All of these provide reference for real estate developers in their project preparation stage.

Comparison of the Methods for Classification of Observations in Predicting Complications in Critically ILL Patients

We compare cluster analysis, Wald's sequential analysis, and the method of coordinate system rotation for the classification of patients with acute destructive pancreatitis and patients with acute combined radiation damage for early detection of complications. In the first case, the Wald analysis is efficient. In the second case, k-means cluster analysis and coordinate rotation method are proposed

Analysis of the Discriminative Methods for Diagnosis of Benign and Malignant Solitary Pulmonary Nodules Based on Serum Markers

Background: Screening indexes of tumor serum markers for benign and malignant solitary pulmonary nodules (SPNs) were analyzed to find the optimum method for diagnosis. Methods: Enzyme-linked immunosorbent assays, an automatic immune analyzer and radioimmunoassay methods were used to examine the levels of 8 serum markers in 164 SPN patients, and the sensitivity for differential diagnosis of malignant or benign SPN was compared for detection using a single plasma marker or a combination of markers. The results for serological indicators that closely relate to benign and malignant SPNs were screened using the Fisher discriminant analysis and a non-conditional logistic regression analysis method, respectively. The results were then verified by the k-means clustering analysis method. Results: The sensitivity when using a combination of serum markers to detect SPN was higher than that using a single marker. By Fisher discriminant analysis, cytokeratin 19 fragments (CYFRA21-1), carbohydrate antigen 125 (CA125), squamous cell carcinoma antigen (SCC) and breast cancer antigen (CA153), which relate to the benign and malignant SPNs, were screened. Through non-conditional logistic regression analysis, CYFRA21-1, SCC and CA153 were obtained. Using the k-means clustering analysis, the cophenetic correlation coefficient (0.940) obtained by the Fisher discriminant analysis was higher than that obtained with logistic regression analysis (0.875). Conclusion: This study indicated that the Fisher discriminant analysis functioned better in screening out serum markers to recognize the benign and malignant SPN. The combined detection of CYFRA21-1, CA125, SCC and CA153 is an effective way to distinguish benign and malignant SPN, and will find an important clinical application in the early diagnosis of SPN.

Exploring structural requirements of aurone derivatives as antimalarials by validated DFT-based QSAR, HQSAR, and COMFA–COMSIA approach

Chloroquine resistance is nowadays a great problem in malaria. Aurone derivatives were effective against chloroquine resistant parasite. Validated density functional theory (DFT)-based chemometric modeling, hologram QSAR (HQSAR), comparative molecular field analysis (CoMFA), and comparative molecular similarity analysis (CoMSIA) studies were conducted on 35 aurone derivatives having antimalarial activity. 2D-QSAR models were developed on the training sets by Y-based ranking method. This model was validated on 50 pairs of the test and the training sets by k-Means cluster analysis method. HQSAR, CoMFA, and CoMSIA models were validated by standard techniques and each method validates the DFT-based 2D-QSAR study and in turn validates the earlier observed structural activity relationship data as well as each other. DFT-based 2D-QSAR model suggests that the increase of Mulliken charge at C14 and HOMO density located on C11 may be conducive to antimalarial activity. Ethyl group attached to C14 and the increase of the value of chemical potential may be beneficial for antimalarial activity. Methoxy fragment is important for better antimalarial activity by HQSAR study. CoMFA analysis shows a favorable steric green region is located near C14 whereas the unfavorable yellow region is far away from C14. A large blue region located near C14 indicates the positively charged groups are favorable at this position. CoMSIA steric features correlates well with the CoMFA steric features. CoMSIA study suggests the bulky hydrophobic substitution at C14 is necessary for antimalarial activity. These results may be utilized to obtain potential antimalarial molecules.

FKM-Based Clustering Case Retrieval of the High-Rise Building Structural Intelligent Scheme Design

Firstly, the disadvantage of the traditional "hard clustering" was analyzed; and the basic principles and algorithm steps of the fuzzy clustering was presented, it can express and process the class of objects which is not clearly in nature. Secondly, the theory and cluster analysis method is introduced to design high-rise structure; based on the fuzzy K-Means clustering analysis, the case retrieval method of the high-level structural intelligent design was established. Lastly, the example of engineering application was given in where the 26 projects data was used as the basis for the cluster and instance retrieval; the former 20 data is used to cluster and the last 4 data used to instance retrieval, the clustering member is k=4. Practice shows that the fuzzy k-means cluster analysis method is effective for the intelligent design of high-rise structure to obtain an instance, and opens up new ways and means for the high-rise building intelligent design.

Passing Stage Division of Bicycle Groups at Signalized Intersection

This paper presents a simple identifying method of bike group ,then takes bicycle groups of signalized intersection as the research object, investigations and video cameras are carried out at two signalized intersections in Beijing and Tianjin city. For the data analysis,by using K-means cluster analysis method ,this paper selects three criterions which are the average speed of bicycle group,the width of bicycle group and the length of bicycle group. Based on three indexes , the passing stage of the bicycle groups at two signalized intersections of Beijing city and Tianjin city are divided into three stages :the first stage for bicycle group gradually spreading ,the second stage for expanding ,the third stage for gradually shrinking.The research on the passing stage of bicycle group starting and running through the intersection at the signalized intersection are useful for determining the capacity of vehicles with the impact of bicycle , and for calculating the signal timing of bicycle under the mixed traffic flow condition at signalized intersection.