• Journal of KIISE
• /
• v.41 no.10
• /
• pp.799-809
• /
• 2014

In this paper, we address the problem of computing Pearson correlation coefficients and Spearman's rank correlation coefficients in a secure manner while data providers preserve privacy of their own data in distributed environment. For a data mining or data analysis in the distributed environment, data providers(data owners) need to share their original data with each other. However, the original data may often contain very sensitive information, and thus, data providers do not prefer to disclose their original data for preserving privacy. In this paper, we formally define the secure correlation computation, SCC in short, as the problem of computing correlation coefficients in the distributed computing environment while preserving the data privacy (i.e., not disclosing the sensitive data) of multiple data providers. We then present SCC solutions for Pearson and Spearman's correlation coefficients using secure scalar product. We show the correctness and secure property of the proposed solutions by presenting theorems and proving them formally. We also empirically show that the proposed solutions can be used for practical applications in the performance aspect.

• IE interfaces
• /
• v.24 no.4
• /
• pp.314-322
• /
• 2011

This study analyzes a relationship between Data Envelopment Analysis(DEA) efficiency scores and a normalization index in order to examine the validity of DEA models. A normalization index concerned in this study is 'sales per R&D project fund' which is regarded as a crucial R&D project performance evaluation index in practice. For this correlation analysis, three distinct DEA models are selected such as DEA basic model, DEA/AR-I revised model(i.e. DEA basic model with Acceptance Region Type I constraints) and Super-Efficiency(SE) model. Especially, SE model is adopted where efficient R&D projects(i.e. Decision Making Units, DMU's) with DEA efficiency score of unity from DEA basic model can be further differentiated in ranks. Considering the non-normality and outliers, two rank correlation coefficients such as Spearman's ${\rho}_s$ and Kendall's ${\tau}_B$ are investigated in addition to Pearson's ${\gamma}$. With an up-to-date empirical massive dataset of n = 482 R&D projects associated with R&D Loan Program of Korea Information Communication Promotion Fund in the year of 2011, statistically significant (+) correlations are verified between the normalization index and every model's DEA efficiency scores with all three correlation coefficients. Especially, the congruence verified in this empirical analysis can be a useful reference for enhancing the practitioner's acceptability onto DEA efficiency scores as a real-world R&D project performance evaluation index.

• Journal of Animal Science and Technology
• /
• v.44 no.5
• /
• pp.523-530
• /
• 2002

Repeatabilities and Correlations among Average Daily Gain, Backfat Thickness and Lean Percent in Swine The repeatability, correlation and rank correlation coefficients among average daily gain, backfat thickness and lean percent were estimated on the basis of records tested from August 1999 to February 2000 with 695 pigs of Duroc, Landrace and Yorkshire boars and gilts tested at 2nd Korea Swine Test Station located in Ha-dong, Kyeongnam Province. The effect of the sex, breed and month of measured were estimated by the least square method. The repeatabilities were estimated from the component of variance among repeated measurements of the trait for the same animal. The results obtained are summarized as follow ; 1. The means of the major economic traits studied were 142.1 days, 173.7 days and 182.5 days for age at 1st, 2nd and 3rd measure, 57.9%, 56.2% and 55.2% for lean percent at 1st, 2nd and 3rd measure, 1.33cm, 1.61cm and 1.63cm for backfat thickness at 1st, 2nd and 3rd measure, 946.6g, 879.2g and 879.4g for average daily gain at 1st, 2nd and 3rd measure, respectively. 2. The correlation coefficients between the backfat thicknesses measured at 1st and 2nd, at 2nd and 3rd, at 1st and 3rd were 0.424, 0.700 and 1.424, respectively. The correlation coefficients between the lean percent measured at 1st and 2nd, at 2nd and 3rd, at 1st and 3rd were 0.493, 0.619 and 0.471, respectively. The correlation coefficients between the average daily gain measured at 1st and 2nd, at 2nd and 3rd, at 1st and 3rd were 0.716, 0.861 and 0.601, respectively. 3. The rank correlation coefficients between backfat thickness measured at 1st and 2nd, at 2nd and 3rd, at 1st and 3rd were 0.438, 0.693 and 0.441, respectively. The rank correlation coefficients between lean percent measured at 1st and 2nd, at 2nd and 3rd, at 1st and 3rd were 0.508, 0.593 and 0.478, respectively. The rank correlation coefficients between average daily gain measured at 1st and 2nd, at 2nd and 3rd, at 1st and 3rd were 0.704, 0.834 and 0.571, respectively. 4. The estimated repeatabilities of the traits studied were 0.428 for the lean percent, 0.374 for the backfat thickness and 0.673 for the average daily gain, respectively.

• Journal of Technologic Dentistry
• /
• v.38 no.3
• /
• pp.217-224
• /
• 2016

Purpose: This study aimed to generate 3-D occlusal curvatures and evaluate the relationship between the occlusal curvatures and mandibular occlusal morphology factors. Methods: Mandibular dental casts from 25 young adult Korean were scanned as a virtual dental models with a 3-D scanner(Scanner S600, Zirkonzahn, Italy). The curve of Spee, curve of Wilson, and Monson's sphere were generated by fitting a circle/sphere to the cusp tips using a least-squares method. The mandibular mesiodistal cusp inclination, buccolingual cusp inclination, and tooth wear parameters were measured on the prepared virtual models using RapidForm2004(INUS technology INC, Seoul, Korea). Wilcoxon signed-rank test was performed to test side difference, and Spearman's rank correlation coefficients were investigated to verify the correlation between occlusal curvatures and correlated factors (a=0.05). Results: The mean radii of curve of Spee were $83.09{\pm}33.94$ in the left side and $79.00{\pm}28.12mm$ in the right side. The mean radii of curve of Wilson were $66.82{\pm}15.87mm$ in the mesial side and $47.87{\pm}9.40mm$ in the distal side with significantly difference between mesiodistal sides(p<0.001). The mean radius of Monson's sphere was $121.85{\pm}47.11mm$. Most of the cusp inclination parameters showed negative correlation for the radius of Monson' sphere(p<0.05). Especially, the buccolingual cusp inclinations in mesial side of molar showed high correlation coefficients among the factors(p<0.05). Conclusion: The radius of Monson's sphere was greater than the classical 4-inch values, and the buccolingual cusp inclinations in mesial side of molar can be considered as one of the main factors correlating with the radius of Monson's sphere.

• Integrative Medicine Research
• /
• v.7 no.2
• /
• pp.136-140
• /
• 2018

Background: The Sasang constitutional analysis tool (SCAT) is an integrated Sasang constitutional analysis system developed by the Korea Institute of Oriental Medicine. This study aimed to evaluate the reliability of a questionnaire for measuring personality and pathophysiological symptoms that is one of the components of the SCAT. Methods: In this study, data were collected from university students in their twenties. Tests were administered twice, with an interval of 4 weeks between tests. Test-retest data from 176 students were collected and used for analysis. Internal consistency reliability was analyzed by using Cronbach's alpha coefficient, and test-retest reliability was analyzed by using Spearman's rank correlation coefficient. Results: Cronbach's alpha coefficient was 0.788 for personality, 0.511 for eating habits, 0.718 for digestion, 0.667 for heat- or cold-wise penchant, and 0.612 for water ingestion. Spearman's rank correlation coefficients, which were used to assess correlations between test and retest results, ranged from 0.444 to 0.828. Conclusion: The internal consistency and test-retest reliability of the SCAT questionnaire were found to be satisfactory.

• Journal of Animal Science and Technology
• /
• v.52 no.2
• /
• pp.97-102
• /
• 2010

The objective of this study was to examine the suitability of genetic evaluation models using the integrated racing records collected from Gwacheon and Busan Gyeongnam racetracks. Results obtained are summarized as follows: In the short-distance races of 1,400 meters and less the records of finishing time at Gwacheon racetrack was superior, whereas, in the races of 1,800 meters and more it was superior in the records from Busan Gyeongnam racetrack. The effects of contemporary groups accounted for 42.7~70.2% of the total variation, and the effects of the individual race considering racing classes was the biggest in all racing distances. Heritabilities and repeatabilities for the finishing time were estimated in the range of 0.153-0.238 and 0.401-0.498, respectively. Correlation coefficients between the breeding values estimated from the integrated records and the breeding values estimated from records of Gwacheon and Busan-Gyeongnam were 0.907 and 0.803, and coefficients of rank correlations were 0.891 and 0.846, respectively. The correlation coefficients between sire's annual earning of the integrated records and Gwacheon and Busan Gyeongnam racetracks records were 0.943 and 0.886, and coefficients of rank correlations were 0.938 and 0.853, respectively. Also, the correlation coefficient of sire's annual earning between Gwacheon and Busan Gyeongnam racetracks was 0.742. The results of this analysis indicate that the genetic evaluation using the integrated racing records are reliable when the racing records from Busan Gyeongnam racetracks are stabilized and more data are accumulated.

• Journal of Preventive Medicine and Public Health
• /
• v.43 no.2
• /
• pp.174-184
• /
• 2010

Objectives: This study shows the issues that should be considered when applying standardized rates using Community Health Survey(CHS) data. Methods: We analyzed 2008 CHS data. In order to obtain the reliability of standardized rates, we calculated z-score and rank correlation coefficients between direct standardized rate and indirect standardized rate for 31 major indices. Especially, we assessed the change of correlations according to population composition (age and sex), and characteristics of the index. We used Mantel-Haenszel chi-square to quantify the difference of population composition. Results: Among 31 major indices, 29 indices' z-score and rank correlation coefficients were over 0.9. However, regions with larger differences in population composition showed lower reliability. Low reliability was also observed for the indices specific to subgroups with small denominator such as 'permanent lesion from stroke', and the index with large regional variations in age-related differences such as 'obtaining health examinations'. Conclusions: Standardized rates may have low reliability, if comparison is made between areas with extremely large differences in population composition, or for indicies with large regional variations in age-related differences. Therefore, the special features of standardized rates should be considered when health state are compared among areas.

• Korean Journal of Community Nutrition
• /
• v.8 no.2
• /
• pp.149-159
• /
• 2003

The purpose of this study was to develop and evaluate the validity of a food frequency questionnaire for Korean adolescents (FFQ-A) which could be used in clinical and epidemiological studies of the lifestyle and health of young people. The FFQ-A was designed to reflect the eating pattern of Korean adolescents, and was based on the 1998 Korean National Health and Nutrition Survey Reports. The FFQ-A had 25 (food categories. A total of 125 subjects (aged 13 to 15 years) was recruited from a randomly chosen middle school in a middle-income neighborhood in Anyang, South Korea. Each subject completed a FFQ-A, as well as a three-day dietary record. Data from 117 subjects (boys 47, girls 70) was used in the final analyses. Data on the nutrients was analyzed to estimate the Pearson correlations, Spearman rank-order correlations and agreement with categories. The validity of the FFQ-A was assessed relative to a three-day dietary record. The Pearson correlation coefficients for all the subjects were 0.94, 0.87, 0.77, 0.79, 0.49 and 0.68 for energy, carbohydrate, protein, fat, calcium, and iron, respectively. Similarly the Spearman rank-order correlation coefficients were 0.94, 0.85, 0.79, 0.81, 0.46, and 0.77 for energy, carbohydrate, protein, fat, calcium and iron, respectively. The Kappa values for energy, carbohydrate, protein, fat, calcium, and iron were 0.88, 0.67, 0.63, 0.67, 0.26, and 0.59, respectively. The percentage for misclassification of the lowest quartile into the highest quartile or vice versa ranged from 0% (energy, carbohydrate, or fat) to 16.7% (Vitamin C). Therefore the FFQ-A has a reasonable ability to assess the energy, carbohydrate, protein and fat intakes as estimated from a three-day dietary record of Korean adolescents. (Korean J Community Nutrition 8(2) : 142∼159, 2003)

• Journal of Astronomy and Space Sciences
• /
• v.23 no.3
• /
• pp.167-176
• /
• 2006

We investigate the correlation between the peak energy and the burst duration using available long GRB data with known redshift, whose circumburst medium type has been suggested via afterglow light curve modeling. We find that the peak energy and the burst duration of the observed GRBs are correlated both in the observer frame and in the GRB rest frame. For our total sample we obtain, for instance, the Spearman rank-order correlation values ${\sim}0.75\;and\;{\sim}0.65$ with the chance probabilities $P=1.0{\times}10^{-3}\;and\;P=6.0{\times}10^{-3}$ in the observer frame and in the GRB rest frame, respectively. We note that taking the effects of the expanding universe into account reduces the value a bit. We further attempt to separate our GRB sample into the 'ISM' GRBs and the 'WIND' GRBs according to environment models inferred from the afterglow light curves and apply statistical tests, as one may expect that clues on the progenitor of GRBs can be deduced directly from prompt emission properties other than from the ambient environment surrounding GRBs. We find that two subsamples of GRBs show different correlation coefficients. That is, the Spearman rank-order correlation are ${\sim}0.65\;and\;{\sim}0.57$ for the 'ISM' GRBs and 'WIND' GRBs, respectively, after taking the effects of the expanding universe into account. It is not yet, however, statistically very much significant that the GRBS in two types of circumburst media show statistically characteristic behaviors, from which one may conclude that all the long bursts are not originated from a single progenitor population. A larger size of data is required to increase the statistical significance.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.38 no.2
• /
• pp.129-137
• /
• 2014

In this study, the characteristics of exhaust gas as a function of the biodiesel mixing ratio were investigated. Diesel and waste oil were used for preparing mixed fuel, and the ratios of the mixed fuel were varied in the BD3~BD100 range. The injection pressures(${\Delta}p_{inj}$) was considered as an experimental variable and was set to 400 bar, 600 bar, 800 bar, 1000 bar, and 1200 bar. Furthermore, for quantitatively analyzing the characteristics of exhaust gas(NOx and Soot), the concepts of Pearson correlation coefficient and Spearman rank-order correlation coefficient based on statistics were introduced. Consequently, it was found that the correlation of the emission of NOx and Soot is linear, and the Pearson and Spearman coefficients are -0.732 and -0.724, respectively, under all analysis conditions. Especially, for the injection pressure of 800 bar, a simultaneous reduction in NOx and Soot emission is possible by controlling the biodiesel mixing ratio. This is because the correlation coefficients of NOx and Soot emissions were nearly 0, as the Pearson correlation coefficient was -0.089.