• Journal of Intelligence and Information Systems
• /
• v.16 no.3
• /
• pp.99-120
• /
• 2010

Due to the wide spread of customers' frequent access of non face-to-face services, there have been many attempts to improve customer satisfaction using huge amounts of data accumulated throughnon face-to-face channels. Usually, a call center is regarded to be one of the most representative non-faced channels. Therefore, it is important that a call center has enough agents to offer high level customer satisfaction. However, managing too many agents would increase the operational costs of a call center by increasing labor costs. Therefore, predicting and calculating the appropriate size of human resources of a call center is one of the most critical success factors of call center management. For this reason, most call centers are currently establishing a department of WFM(Work Force Management) to estimate the appropriate number of agents and to direct much effort to predict the volume of inbound calls. In real world applications, inbound call prediction is usually performed based on the intuition and experience of a domain expert. In other words, a domain expert usually predicts the volume of calls by calculating the average call of some periods and adjusting the average according tohis/her subjective estimation. However, this kind of approach has radical limitations in that the result of prediction might be strongly affected by the expert's personal experience and competence. It is often the case that a domain expert may predict inbound calls quite differently from anotherif the two experts have mutually different opinions on selecting influential variables and priorities among the variables. Moreover, it is almost impossible to logically clarify the process of expert's subjective prediction. Currently, to overcome the limitations of subjective call prediction, most call centers are adopting a WFMS(Workforce Management System) package in which expert's best practices are systemized. With WFMS, a user can predict the volume of calls by calculating the average call of each day of the week, excluding some eventful days. However, WFMS costs too much capital during the early stage of system establishment. Moreover, it is hard to reflect new information ontothe system when some factors affecting the amount of calls have been changed. In this paper, we attempt to devise a new model for predicting inbound calls that is not only based on theoretical background but also easily applicable to real world applications. Our model was mainly developed by the interactive decision tree technique, one of the most popular techniques in data mining. Therefore, we expect that our model can predict inbound calls automatically based on historical data, and it can utilize expert's domain knowledge during the process of tree construction. To analyze the accuracy of our model, we performed intensive experiments on a real case of one of the largest car insurance companies in Korea. In the case study, the prediction accuracy of the devised two models and traditional WFMS are analyzed with respect to the various error rates allowable. The experiments reveal that our data mining-based two models outperform WFMS in terms of predicting the amount of accident calls and fault calls in most experimental situations examined.

• Asian Pacific Journal of Cancer Prevention
• /
• v.15 no.22
• /
• pp.10003-10008
• /
• 2014

Background: Cervical cancer has been a leading female cancer in Thailand for decades, and has been second to breast cancer after 2007. The Ministry of Public Health (MoPH) has provided opportunistic screening with Pap smears for more than 30 years. In 2002, the MoPH and the National Health Security Office provided countrywide systematic screening of cervical cancer to all Thai women aged 35-60 years under universal health care coverage insurance scheme at 5-year intervals. Objectives: This study characterized the cervical cancer incidence trends in Songkhla in southern Thailand using joinpoint and age period cohort (APC) analysis to observe the effect of cervical cancer screening activities in the past decades, and to project cervical cancer rates in the province, to 2030. Materials and Methods: Invasive and in situ cervical cancer cases were extracted from the Songkhla Cancer Registry from 1990 through 2010. Age standardized incidence rates were estimated. Trends in incidences were evaluated by joinpoint and APC regression models. The Norpred package was modified for R and was used to project the future trends to 2030 using the power of 5 function and cut trend method. Results: Cervical cancer incidence in Songkhla peaked around 1998-2000 and then dropped by -4.7% per year. APC analysis demonstrated that in situ tumors caused an increase in incidence in early ages, younger cohorts, and in later years of diagnosis. Conclusions: Both joinpoint and APC analysis give the same conclusion in continuation of a declining trend of cervical cancer to 2030 but with different rates and the predicted goal of ASR below 10 or even 5 per 100,000 women by 2030 would be achieved. Thus, maintenance and improvement of the screening program should be continued. Other population based cancer registries in Thailand should analyze their data to confirm the success of cervical cancer screening policy of Thailand.

• Journal of Korea Water Resources Association
• /
• v.54 no.12
• /
• pp.1205-1214
• /
• 2021

Precipitation is one of a major causes of landslides by rising of pore water pressure, which leads to fluctuations of soil strength and stress. For this reason, precipitation is the most frequently used to determine the landslide thresholds. However, using only precipitation has limitations in predicting and estimating slope stability quantitatively for reducing false alarm events. On the other hand, Soil Moisture (SM) has been used for calculating slope stability in many studies since it is directly related to pore water pressure than precipitation. Therefore, this study attempted to evaluate the appropriateness of applying soil moisture in determining the landslide threshold. First, the reactivity of soil saturation level to precipitation was identified through time-series analysis. The precipitation threshold was calculated using daily precipitation (P_daily) and the Antecedent Precipitation Index (API), and the hydrological threshold was calculated using daily precipitation and soil saturation level. Using a contingency table, these two thresholds were assessed qualitatively. In results, compared to P_daily only threshold, Goesan showed an improvement of 75% (P_daily + API) and 42% (P_daily + SM) and Changsu showed an improvement of 33% (P_daily + API) and 44% (P_daily + SM), respectively. Both API and SM effectively enhanced the Critical Success Index (CSI) and reduced the False Alarm Rate (FAR). In the future, studies such as calculating rainfall intensity required to cause/trigger landslides through soil saturation level or estimating rainfall resistance according to the soil saturation level are expected to contribute to improving landslide prediction accuracy.

• Journal of Korea Water Resources Association
• /
• v.52 no.7
• /
• pp.475-482
• /
• 2019

Subseasonal-to-Seasonal (S2S) prediction information which have 2 weeks to 2 months lead time are expected to be used through many parts of industry fields, but utilizability is not reached to expectation because of lower predictability than weather forecast and mid- /long-term forecast. In this study, we used multi-layer perceptron (MLP) which is one of machine learning technique that was built for regression training in order to improve predictability of S2S precipitation data at South Korea through post-processing. Hindcast information of ECMWF was used for MLP training and the original data were compared with trained outputs based on dichotomous forecast technique. As a result, Bias score, accuracy, and Critical Success Index (CSI) of trained output were improved on average by 59.7%, 124.3% and 88.5%, respectively. Probability of detection (POD) score was decreased on average by 9.5% and the reason was analyzed that ECMWF's model excessively predicted precipitation days. In this study, we confirmed that predictability of ECMWF's S2S information can be improved by post-processing using MLP even the predictability of original data was low. The results of this study can be used to increase the capability of S2S information in water resource and agricultural fields.

• Journal of Korean Academy of Nursing
• /
• v.53 no.4
• /
• pp.468-479
• /
• 2023

Purpose: This study aimed to evaluate the use of the respiratory rate oxygenation (ROX) index, ROX-heart rate (ROX-HR) index, and saturation of percutaneous oxygen/fraction of inspired oxygen ratio (SF ratio) to predict weaning from high-flow nasal cannula (HFNC) in patients with respiratory distress in a pediatric intensive care unit. Methods: A total of 107 children admitted to the pediatric intensive care unit were enrolled in the study between January 1, 2017, and December 31, 2021. Data on clinical and personal information, ROX index, ROX-HR index, and SF ratio were collected from nursing records. The data were analyzed using an independent t-test, χ² test, Mann-Whitney U test, and area under the curve (AUC). Results: Seventy-five (70.1%) patients were successfully weaned from HFNC, while 32 (29.9%) failed. Considering specificity and sensitivity, the optimal cut off points for predicting treatment success and failure of HFNC oxygen therapy were 6.88 and 10.16 (ROX index), 5.23 and 8.61 (ROX-HR index), and 198.75 and 353.15 (SF ratio), respectively. The measurement of time showed that the most significant AUC was 1 hour before HFNC interruption. Conclusion: The ROX index, ROX-HR index, and SF ratio appear to be promising tools for the early prediction of treatment success or failure in patients initiated on HFNC for acute hypoxemic respiratory failure. Nurses caring for critically ill pediatric patients should closely observe and periodically check their breathing patterns. It is important to continuously monitor three indexes to ensure that ventilation assistance therapy is started at the right time.

• Asia-pacific Journal of Multimedia Services Convergent with Art, Humanities, and Sociology
• /
• v.7 no.4
• /
• pp.593-601
• /
• 2017

In this paper, we constructed the model of production/investment, distribution, and screening by using variables that can be considered at each stage according to the value chain stage of the movie industry. To increase the predictive power of the model, a regression analysis was used to derive meaningful variables. Based on the given variables, we compared the difference in predictive power between the artificial neural network, which is a machine learning analysis method, and the decision tree analysis method. As a result, the accuracy of artificial neural network was higher than that of decision trees when all variables were added in production/ investment model and distribution model. However, decision trees were more accurate when selected variables were applied according to regression analysis results. In the screening model, the accuracy of the artificial neural network was higher than the accuracy of the decision tree regardless of whether the regression analysis result was reflected or not. This paper has an implication which we tried to improve the performance of movie prediction model by using machine learning analysis. In addition, we tried to overcome a limitation of linear approach by reflecting the results of regression analysis to ANN and decision tree model.

• The Journal of Society for e-Business Studies
• /
• v.21 no.1
• /
• pp.15-32
• /
• 2016

Online news on the Internet, as published open data, contain facts or opinions about a specific affair and hence influences considerably on the decisions of the general publics who are interested in a particular issue. Therefore, we can predict the people's choices related with the issue by analyzing a large number of related internet news. This study aims to propose a text analysis methodto predict the outcomes of events that take place in a specific place. We used topics of the news articles because the topics contains more essential text than the news articles. Moreover, when it comes to mobile environment, people tend to rely more on the news topics before clicking into the news articles. We collected the titles of news articles and divided them into the learning and evaluation data set. Morphemes are extracted and their polarity values are identified with the learning data. Then we analyzed the sensitivity of the entire articles. As a result, the prediction success rate was 70.6% and it showed a clear difference with other analytical methods to compare. Derived prediction information will be helpful in determining the expected demand of goods when preparing the event.

• Korean Journal of Construction Engineering and Management
• /
• v.16 no.2
• /
• pp.65-76
• /
• 2015

The Clean Development Mechanism (CDM) is the multi-lateral 'cap and trade' system endorsed by the Kyoto Protocol. CDM allows developed (Annex I) countries to buy CER credits from New and Renewable (NE) projects of non-Annex countries, to meet their carbon reduction requirements. This in effect subsidizes and promotes NE projects in developing countries, ultimately reducing global greenhouse gases (GHG). To be registered as a CDM project, the project must prove 'additionality,' which depends on numerous factors including the adopted technology, baseline methodology, emission reductions, and the project's internal rate of return. This makes it difficult to determine ex ante a project's acceptance as a CDM approved project, and entails sunk costs and even project cancellation to its project stakeholders. Focusing on hydro power projects and employing UNFCCC public data, this research developed a prediction model using logistic regression and CART to determine the likelihood of approval as a CDM project. The AUC for the logistic regression and CART model was 0.7674 and 0.7231 respectively, which proves the model's prediction accuracy. More importantly, results indicate that the emission reduction amount, MW per hour, investment/Emission as crucial variables, whereas the baseline methodology and technology types were insignificant. This demonstrates that at least for hydro power projects, the specific technology is not as important as the amount of emission reductions and relatively small scale projects and investment to carbon reduction ratios.

• The Korean Journal of Applied Statistics
• /
• v.28 no.5
• /
• pp.977-990
• /
• 2015

Food delivery services are well developed in the Republic of Korea, The increase of one person households and the success of app applications influence delivery services these days. We consider a prediction model for the food delivery service based on weather and dates to predict the number of food delivery services in 2014 using various data mining techniques. We use linear regression, random forest, gradient boosting, support vector machines, neural networks, and logistic regression to find the best prediction model. There are four categories of food delivery services and we consider two methods. For the first method, we estimate the total number of delivery services and the posterior probabilities of each delivery service. For the second method, we use different models for each category and combine them to estimate the total number of delivery services. The neural network and linear regression model perform best in the first method, this is followed by the neural network which is the best for the second method. The result shows that we can estimate the number of deliveries accurately based on dates and weather information.

• Journal of the Korea Convergence Society
• /
• v.10 no.4
• /
• pp.1-6
• /
• 2019

Neural networks have been used to predict the direction of stock index movement from past data. The conventional research that predicts the upward or downward movement of the stock index predicts a rise or fall even with small changes in the index. It is highly likely that losses will occur when trading ETFs by use of the prediction. In this paper, a neural network model that predicts the movement direction of the daily KOrea composite Stock Price Index (KOSPI) to reduce ETF trading losses and earn more than a certain amount per trading is presented. The proposed model has outputs that represent rising (change rate in index ${\geq}{\alpha}$), falling (change rate ${\leq}-{\alpha}$) and neutral ($-{\alpha}$ change rate < ${\alpha}$). If the forecast is rising, buy the Leveraged Exchange Traded Fund (ETF); if it is falling, buy the inverse ETF. The hit ratio (HR) of PNN1 implemented in this paper is 0.720 and 0.616 in the learning and the evaluation respectively. ETF trading yields a yield of 8.386 to 16.324 %. The proposed models show the better ETF trading success rate and yield than the neural network models predicting KOSPI.