• Journal of Korea Society of Digital Industry and Information Management
• /
• v.15 no.3
• /
• pp.197-206
• /
• 2019

Telemarketing has been used in finance due to the reduction of offline channels. In order to select telemarketing target customers, various machine learning techniques have emerged to maximize the effect of minimum cost. However, there are problems that the class imbalance, which the number of marketing success customers is smaller than the number of failed customers, and the recall rate is lower than accuracy. In this paper, we propose a method that solve the imbalanced class problem and increase the recall rate to improve the efficiency. The hybrid sampling method is applied to balance the data in the class, and the stacked deep network is applied to improve the recall and precision as well as the accuracy. The proposed method is applied to actual bank telemarketing data. As a result of the comparison experiment, the accuracy, the recall, and the precision is improved higher than that of the conventional methods.

• Particle and aerosol research
• /
• v.19 no.3
• /
• pp.77-87
• /
• 2023

We present the development of a wetcyclone sampler designed for the sampling of airborne bacteria. The wetcyclone design involves a combination of two traditional cyclone shapes and computational fluid dynamics (CFD) analysis to validate its effectiveness in terms of pressure drop and collection efficiency. The wetcyclone exhibits a collection efficiency of over 90% for bacteria, specifically targeting Staphylococcus aureus. Additionally, the wetcyclone enables continuous bioaerosol sampling using a liquid medium (deionized water), demonstrating a concentration ratio exceeding >105 and a stable microbial recovery rate of 81.9%. The application of real-time quantitative polymerase chain reaction (qPCR) and the colony counting method ensures precise measurement of the concentration ratio and microbial recovery rate.

• Journal of Animal Science and Technology
• /
• v.48 no.4
• /
• pp.541-554
• /
• 2006

Present review is to introduce an omasal sampling technique in rumen proteolysis and to consider some information on the omasal sampling technique with particular emphasis on methodological aspects. Use of the omasal sampling technique provides a new opportunity for accurate estimation of rumen metabolism with overcoming limitations of previous in vivo, in vitro and/or in situ methods. The potential advantages of the present technique compared with post-ruminal sampling techniques include following points; 1) only rumen cannulated animals are required, 2) less endogenous nitrogen (N) is contaminated in omasal digesta and 3) omasal digesta are devoid of exposure to acid peptide hydrolysis occurring in the abomasum. Estimates of soluble non-ammonia N (SNAN) in omasal digesta indicate that the assumptions underlying the in situ method that rapidly degradable N fraction can be degraded at an infinite rate and only insoluble dietary N escapes the rumen may be not valid. Quatitatively higher peptide concentration rather than free amino acid and soluble protein in escapable SNAN suggests that hydrolysis of peptide to amino acid may be the rate-limiting step in rumen proteolysis.

• Journal of Korea Spatial Information System Society
• /
• v.11 no.2
• /
• pp.89-98
• /
• 2009

As data stream is entered into system continuously and the memory space is limited, the data exceeding the memory size cannot be processed. In order to solve the problem, load shedding methods which drop a part of data to prevent exceeding the storage space have been researched. Generally, a traditional load shedding method uses random sampling with optimized rate according to data deviation. The method samples data not to distinguish those used in spatial query because the method uses only a random sampling with optimized rate according to data deviation. Therefore, the accuracy of query was reduced in u-GIS environment including spatial query. In this paper, we researched a new load shedding method improving accuracy of the query in u-GIS environment which runs spatial query and aspatial query simultaneously. The method uses a new sampling method that samples data having low probability used in query. Therefore proposed method improves spatial query accuracy and query processing speed as applying spatial filtering operation to sampling operator.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.23 no.9
• /
• pp.1151-1162
• /
• 1999

The effects of radial heat and $H_2O$ diffusion on the evolution of silica particles in coflow diffusion flames have been studied experimentally. The evolution of silica aggregate particles in coflow diffusion flames has been measured experimentally using light scattering and thermophoretic sampling techniques. The measurements of scattering cross section from $90^{\circ}$ light scattering have been utilized to calculate the aggregate number density and volume fraction using with combination of measuring the particle size and morphology through the localized sampling and a TEM image analysis. Aggregate or particle number densities and volume fractions were calculated using Rayleigh-Debye-Gans and Mie theory for fractal aggregates and spherical particles, respectively. Flame temperatures and volumetric differential scattering cross sections have been measured for different flame conditions such as inert gas species, $H_2$ flow rates, and burner injection configurations to examine the relation between the formation of particles and radial $H_2O$ diffusion. The comparisons of oxidation and flame hydrolysis have also been made for various $H_2$ flow rates using $N_2$ or $O_2$ as a carrier gas. Results indicate that the role of oxidation becomes dominant as both carrier gas($O_2$) and $H_2$ flow rates increases since the radial heat diffusion precedes $H_2O$ diffusion in coflow flames used in this study. The effect of carrier gas flow rates on the evolution of silica particles have also been studied. When using $N_2$ as a carrier gas, the particle volume fraction has a maximum at a certain carrier gas flow rate and as the flow rate is further increased, the hydrolysis reaction Is delayed and the spherical particles finally evolves into fractal aggregates due to decreased flame temperature and residence time.

• Journal of Bio-Environment Control
• /
• v.6 no.4
• /
• pp.310-316
• /
• 1997

This experiments comparisons were conducted to establish the effective sap analytical method including quantitative analysis of nitrate and phosphorous, sap extraction method, extraction rate, macerating time and sampling position. Earl's Favorite muskmelon was sown on February 11th, 1993 and transplanted to rockwool slab when three to fourth true leaf appeared. Non-circulating system was employed in this experiment. Sap testing using a sample provides an immediate analysis of nutritional status of the crop and potential problems can be recognized and corrective measures taken. Some methods were recommended UV-spectroscopy for the analysis of nitrate nitrogen, Vanad molybdenium acid method for phosphorous. Optimal extraction rate of sample to water was determined as one to four and the best result observed at 60seconds in macerating duration. Further trials would be requested about sampling time due to the difference according to the sampling time. Given sampling position was favorable for the sap analysis.

• Journal of The Korean Society of Agricultural Engineers
• /
• v.50 no.2
• /
• pp.65-71
• /
• 2008

A monsoon season monitoring data from June to September, 2005 of a small forested watershed located at the upstream of the North Han River system in Korea was conducted to analyze the flow variations, the NPS pollutant concentrations, and the pollution load characteristics with respect to sampling frequencies. During the 4-month period, 1,423 mm or 79.2% of annual rainfall(1,797 mm) were occurred and more than 77%, 54% and 68% of annual T-N, $NO_3$-N and T-P loads discharged. Flow rate was continuously measured with automatic velocity and water level meters and 58 water quality samples were taken and analyzed. It was analyzed that the flow volume by random measurement varied very widely and ranged from 79% to 218% of that of continuous measurement. It was recommended that flow measurement of small forested watersheds should be continuously measured with automated flow meters to precisely measure flow rates. Flow-weighted mean concentrations of T-N, $NO_3$-N and T-P during the period were 2.114 mg/L, 0.836 mg/L, and 0.136 mg/L, respectively. T-N, $NO_3$-N and T-P loads were sensitive to the number of samples. And it was analyzed that in order to measure the pollution load within the error of 10% to the true load, the rate of sampling frequency should be higher than 89.7% of the sample numbers that were required to compute the true pollution load. If it is compared to selected foreign research results, about 10 water samples for each rainfall event were needed to compute the pollution load within 10% error. It is unlikely in Korea and recommended that thorough NPS pollution monitoring studies are required to develop the standard monitoring procedures for reliable NPS pollution quantification.

• The Research Journal of the Costume Culture
• /
• v.17 no.6
• /
• pp.1141-1154
• /
• 2009

As a preliminary research of body characteristics analysis of Korean obese woman, this study aims to select optimal obesity judgment tools for reliable sampling of obese subjects from 2,425 female measurement data out of 2004 Size Korea project's raw data. From previous researches related to obesity, 7 obesity judgment tools were chosen. 2007 obesity rate(26.3%) of Korean female adults was refered in selecting optimal obesity judgment criteria in the study. The results are as follows. Firstly, it was verified that BMI was the most suitable in judging and sampling the obese subjects by the percentile analysis. R$\ddot{o}$hrer index was also reliable in grouping the obese subjects from a population. Secondly, it was concluded that the obesity ratios of relative weight 120 and higher group, R$\ddot{o}$hrer index 1.6 and higher group and waist girth 80cm and higher group were the most similar to obesity rate of Korean female adults by 2007 National Health & Nutrition Survey. Thirdly, 30 direct measurements, age, 2 drop values and 6 ratios of 7 groups by the obesity judgment tools showed the significance each other at p<0.001 level. On the other hand, "bust point to bust point" and "waist to hip length" measurements didn't show the significant differences among 7 groups. Conclusively, 4 to 5 satisfactions out of 7 obesity judgment criteria were adequate and sufficient in sampling the obese subjects. If it is needed the strict criteria for judging the obesity, 5 satisfactions and higher group will be the best choice as the obese subjects. However 4 satisfactions and higher group generally, will be adequate for sampling of the obese subjects.

• Proceedings of the Korean Society of Life Science Conference
• /
• 2007.10a
• /
• pp.82.2-82.2
• /
• 2007

See Full Text

• Journal of the military operations research society of Korea
• /
• v.11 no.2
• /
• pp.64-68
• /
• 1985

This paper considers the following case: (1) the product is paid by the right price for a lot accepted by a given consumer's acceptance sampling plan, and (2) the product is paid by the discounted price for a lot rejected by this plan. In such a case, the producer's sampling plan need not be the same as that of the consumer's. From the producer's view point, the producer need to determine the preliminary sample size which maximizes his revenue. This paper, therefore, determines an optimal preliminary sample size from the producer's view point. This preliminary sample size is affected by the consumer's acceptance sampling plan, percent defective, preliminary inspection cost and the discount rate of the price.