• Journal of the Korean Data and Information Science Society
• /
• 제19권4호
• /
• pp.1007-1018
• /
• 2008

The purpose of this case study is to demonstrate database-marketing management. First, we explore original variables for insurance customer's data, modify them if necessary, and go through variable selection process before analysis. Then, we develop churn prediction models using logistic regression, neural network and SVM analysis. We also compare these three data mining models in terms of misclassification rate.

• 디지털산업정보학회논문지
• /
• 제20권2호
• /
• pp.1-10
• /
• 2024

For the purpose of predicting credit card customer churn accurately through data analysis, a model can be constructed with various machine learning algorithms, including decision tree. And feature importance has been utilized in selecting better input features that can improve performance of data analysis models for several application areas. In this paper, a method of utilizing feature importance calculated from the MDI method and its effects are investigated in the credit card customer churn prediction problem with classification trees. Compared with several random feature selections from case data, a set of input features selected from higher value of feature importance shows higher predictive power. It can be an efficient method for classifying and choosing input features necessary for improving prediction performance. The method organized in this paper can be an alternative to the selection of input features using feature importance in composing and using classification trees, including credit card customer churn prediction.

• Journal of Welding and Joining
• /
• 제31권5호
• /
• pp.7-14
• /
• 2013

The legitimacy of dominating factor in the high accuracy prediction of welding distortion was investigated for butt welding and fillet welding. When out-of-plane distortion was measured by the experiment objecting to butt welding, if tack welding was easily performed, the position of a neutral axis was variously changed by the irregularity. Then, there have been a case that out-of-plane distortion was generated in the unexpected direction. This case should be especially noted. New model for the experiment was proposed so as to solve this problem. As it was elucidated by the case of fillet welding, it was verified that the analysis should be carried out with satisfying the yield condition (especially at high temperature above 700 degree Celsius) and with closely simulating the penetration shape (heat input in weld metal) in order to solve the proposition that is the high accuracy prediction of welding distortion. It was confirmed that residual stress is highly predicted because welding distortion is highly predicted, too.

• International Journal of Air-Conditioning and Refrigeration
• /
• 제13권2호
• /
• pp.89-98
• /
• 2005

In this study, indoor VOCs concentration emitted from floor and furniture was measured after the installation of floor and furniture in a real residence. With the measured data, prediction method and predication equations for indoor concentration of each VOCs and BTEX were developed. The following conclusions were drawn from this study. First, according to the predicted results of concentration decrease of BTEX (benzene, toluene, ethylbenzene, m,p,o-xylene) after the installation of floor in a real residence, prediction equation can be expressed using exponential function. Second, in case of floor, more reliable prediction equation can be obtained by using cumulative value of indoor concentration than by using just hourly measured value directly. Indoor concentration of benzene can be expressed as $y=408.52(1­e^{-00031{\times}time})$ with $R^2$ of 0.94 which is significantly high value. Third, toluene showed the highest concentration in case of furniture installation indoors, and it needed the longest time for concentration decrease. However, other substances except toluene showed constant concentration throughout the measurement period. Fourth, in case of furniture installation indoors, prediction equation of toluene concentration decrease is estimated to be $y= 3616.3{\times}e^{(-0.1091{\times}time)}+513.96{\times}e^{(-0.0006{\times}time)}\;with\; R^2$ of 0.95 which is significantly high value.

• 대기
• /
• 제27권4호
• /
• pp.385-398
• /
• 2017

Numerical weather prediction experiments were carried out by applying topographic effects to reduce or enhance the solar radiation by terrain. In this study, x and ${\kappa}({\phi}_o,\;{\theta}_o)$ are precalculated for topographic effect on high resolution numerical weather prediction (NWP) with 1 km spatial resolution, and meteorological variables are analyzed through the numerical experiments. For the numerical simulations, cases were selected in winter (CASE 1) and summer (CASE 2). In the CASE 2, topographic effect was observed on the southward surface to enhance the solar energy reaching the surface, and enhance surface temperature and temperature at 2 m. Especially, the surface temperature is changed sensitively due to the change of the solar energy on the surface, but the change of the precipitation is difficult to match of topographic effect. As a result of the verification using Korea Meteorological Administration (KMA) Automated Weather System (AWS) data on Seoul metropolitan area, the topographic effect is very weak in the winter case. In the CASE 1, the improvement of accuracy was numerically confirmed by decreasing the bias and RMSE (Root mean square error) of temperature at 2 m, wind speed at 10 m and relative humidity. However, the accuracy of rainfall prediction (Threat score (TS), BIAS, equitable threat score (ETS)) with topographic effect is decreased compared to without topographic effect. It is analyzed that the topographic effect improves the solar radiation on surface and affect the enhancements of surface temperature, 2 meter temperature, wind speed, and PBL height.

• 한국지반공학회지:지반
• /
• 제12권5호
• /
• pp.103-116
• /
• 1996

교통하중이나 건설 작업에 의해 야기되는 지반진동은 인접구조물에 손상을 입힐 수 있으므로 중요하게 관리되어야 한다. 최근 문헌에 의하면 도심지에서 구조물 손상의 주요 원인은 진동파의 직접 전달에 의한 손상보다 지반다짐으로 야기되는 부등침하에 의한 것으로 알려져 있다. 덕 논문에서는 지반진동에 의한 주변침하 예측기 법을 사례 연구를 충심으로 소개하였다. 실험실에서 개발된 침하예측모델의 현장적용기법을 설명하였으며, 사례연구에서 계측된 침하량과 모델에서 예측된 침하량을 비교하였다. 모델에서 예측한 침하량은 실제침하량과 비교적 잘 일치하였다. 전형적인 도심지 상황에서 영향요소 연구를 수행하여 침하예민요소들을 살펴보고 현장에서의 신뢰성있는 진동 계측 및 평가 방법을 제안하였다. 이상의 연구를 통해 사직토 지반에서 진동에 의한 침하평가 및 예측을 위해 개발된 모델의 사용 가능성을 입증하였다.

• 한국분무공학회지
• /
• 제25권4호
• /
• pp.188-195
• /
• 2020

To meet stringent emission regulations of automotive engines, fuel injection control techniques have advanced based on reliable and fast computing prediction models. This study aims to develop a reliable lightweight prediction model of fuel injection rates using a small number of input parameters and based on simple fluid dynamic theories. The prediction model uses the geometry of the injector nozzle, needle motion data, injection conditions and the fuel properties. A commercial diesel injector and US No. 2 diesel were used as the test injector and fuel, respectively. The needle motion data were measured using X-ray phase-contrast imaging technique under various fuel injection pressures and injection pulse durations. The actual injector rate profiles were measured using an injection rate meter for the validation of the model prediction results. In the case of long injection durations with the steady-state operation, the model prediction results showed over 99 % consistency with the measurement results. However, in the case of short injection cases with the transient operation, the prediction model overestimated the injection rate that needs to be further improved.

• 한국지능시스템학회:학술대회논문집
• /
• 한국지능시스템학회 2007년도 추계학술대회 학술발표 논문집
• /
• pp.283-286
• /
• 2007

This paper proposes a genetic algorithm (GA) approach to instance selection in case-based reasoning (CBR) for the prediction of Korea Stock Price Index (KOSPI). CBR has been widely used in various areas because of its convenience and strength in complex problem solving. Nonetheless, compared to other machine learning techniques, CBR has been criticized because of its low prediction accuracy. Generally, in order to obtain successful results from CBR, effective retrieval of useful prior cases for the given problem is essential. However, designing a good matching and retrieval mechanism for CBR systems is still a controversial research issue. In this paper, the GA optimizes simultaneously feature weights and a selection task for relevant instances for achieving good matching and retrieval in a CBR system. This study applies the proposed model to stock market analysis. Experimental results show that the GA approach is a promising method for instance selection in CBR.

• 한국항공운항학회지
• /
• 제17권4호
• /
• pp.25-31
• /
• 2009

In development phase of the Aircraft Systems, the reliability prediction of electronic equipments are usually performed using MIL-HDBK-217. The reliability of fielded electronic systems, however, used to be underestimated with MIL-HDBK-217. To solve this problem, some alternatives are suggested and Telcordia SR-332 is among them. In this case study, the reliability of ESU which controls gas turbine engine is predicted using Telcordia SR-332 along with the development test data. The predicted reliabilities of ESU using Telcordia SR-332 and MIL-HDBK-217 are also compared. As a result this case study showed that the predicted reliability using Telcordia SR-332 was better close to field(operation) reliability than MIL-HDBK-217.

• 환경위생공학
• /
• 제23권2호
• /
• pp.19-25
• /
• 2008

For measurement of tylosin, ammonia nitrogen, and glucose concentration during the culture of Streptomyces fradiae using Near Infrared Spectroscopy, the calibration using various mathematical models was performed and then, based on the linear model, the validation was carried out. In the case of sucrose concentration using the MLR method, the Standard Error of Prediction and Multiple correlation coefficient were 1.97, and 0.991, respectively. In the case of ammonia nitrogen concentration using the PLSR method, the Standard Error of Prediction and Multiple correlation coefficient were 0.13, and 0.990, respectively. In the case of tylosin concentration using the PLSR method, the standard Error of Prediction and Multiple correlation coefficient were 0.54, and 0.984, respectively.