• Communications for Statistical Applications and Methods
• /
• 제11권2호
• /
• pp.227-233
• /
• 2004

This paper presents our work on constructing a model that is intended to predict the probability of attrition at specified points in time among customers of an insurance company. There are some difficulties in building a data-based model because a data set may contain possibly censored observations. In an effort to avoid such kind of problem, we performed logistic regression over specified time intervals while using explanatory variables to construct the proposed model. Then, we developed a Cox-type regression model for estimating the probability of attrition over a specified period of time using time-dependent explanatory variables subject to changes in value over the course of the observations.

• Communications for Statistical Applications and Methods
• /
• 제9권1호
• /
• pp.271-280
• /
• 2002

Recently, many companies have applied data mining techniques to promote competitive power in the field of their business market. In this study, we address how data mining, that is a technique to enable to discover knowledge from a deluge of data, Is used in an executed project in order to support decision making of an enterprise. Also, we develope scoring model on customer attrition probability for automobile-insurance company using data mining techniques. The development of scoring model in domestic insurance is given as an example concretely.

• 한국국방경영분석학회지
• /
• 제28권2호
• /
• pp.70-84
• /
• 2002

This paper deals with the attrition rate of major facility such as a particular building that is one of the most important target in the war time. In order to estimate the attrition rate, we use JAWS, WinJMEM which are programed by JTCG/ME of AMSAA and spreadsheet package which is able to assist the limitation of those programs and calculate all the procedure of this computation. This method uses the effectiveness index(El) which indicates the numerical measure of the effectiveness of a given weapon of a given target. The range error probable(REP) and the deflection error probable(DEP) in the ground plane also should be used. Those mean the measure of delivery accuracy of the weapon system. In this paper, it is improved that the El can be obtained from the regression analysis using the weight of the warhead explosive as the independent variable. It implies that we are able to obtain the El and the conditional probability of damage of the enemy weapon. After that, the single-sortie probability of damage can be computed using WinJMEM or another assistant program such as the spreadsheet package which shows the result immediately.

• 한국국방경영분석학회지
• /
• 제12권2호
• /
• pp.22-36
• /
• 1986

The study is to design optimal aircraft allocation model for sufficing the required level of damage, minimizing attrition cost when the aircrafts attack the enemy's fixed target. When friendly aircraft attacks enemy target, the aircraft will suffer the loss due to the enemy's anti-aircraft weapons and aircraft. For this study, it is required that the probability of target damage by the type of aircraft, level of target damage and attrition cost are computed for the application of this model.

• 한국국방경영분석학회지
• /
• 제3권2호
• /
• pp.119-134
• /
• 1977

The goal of this study is to determine which is the better bomb; a general purpose bomb or a special bomb, from the point of view of cost-effectiveness of air munitions. What we call the special bomb is a laser guided bomb which adds a guidance control unit on the head of a general purpose bomb. Althrough their characteristics of explosion are the same, their effectivenesses are very defferent due to their probability of hitting the targets. Now, in order to measure their relative a effectiveness, we could select various enemy targets such as runway, radar, and then Ive could also calculate requirments of both bombs and firghter aircraft sorties according to the desired damage level of each target, Following above steps, we can derive the total costs and compare effectiveness. As a results of this study, when the attrition rate of aircraft is under $1.1\%$, the general purpose bomb is better than the laser guided bomb. Above $1.1\%$ attrition rate, we could say that luser guided bomb is the most economical bomb.

• 대한교통학회지
• /
• 제18권6호
• /
• pp.63-75
• /
• 2000

SP 데이터는 데이터 수집의 효율이 RP 데이터 보다 높고 장래의 교통 시스템의 조건이나 속성에 대한 응답자들의 태도를 조사 할 수 있다는 점에서 많이 사용되고 있으나 SP 데이터는 주요하게 두 가지 편위를 가지고 있는데 SP 설문조사시에 발생하는 응답편위와 SP 패널조사시에 발생하는 누락편위이다. 이러한 SP 데이터의 편위가 수정되지 않으면 장래의 잘못된 교통수요예측을 유발할 수 있다. 따라서 본 연구에서는 이러한 SP 모델의 편위와 상태의존을 고려한 모델을 구축하기 위하여 6개의 횡단면 모델과 동적모델을 제안하였다. 횡단면 모델 중 RP데이터의 선택결과를 고려한 모델을 이용하여 SP모델의 편위를 보완할 수 있는 모델을 구축할 수 있었으며 동적모델의 경우에 패널데이터의 상태의존도를 지수함수로 가정하여 상태의존도를 고려한 동적모델을 구축하였다. 또한 패널조사시에 필연적으로 발생하는 누락데이터에 의한 누락편위를 모델에 고려하기 위하여 WESML방법을 적용하여 모델을 구축하였으며 그 결과 상태의존도를 보다 세밀하게 제어함으로서 모델의 설명력을 개선시키고 향후 SP 패널데이터를 이용한 동적모델의 적용성을 높일 수 있음을 알 수 있었다. 본 연구에서는 모델의 유용성을 검토하기 위하여 전주시의 외각 지역인 호남제일문 방향에서 도심으로 접근하는 3개의 주경로(천변로, 기린로, 팔달로)에 대한 패널조사 자료를 바탕으로 모델을 구축하였다.

• Journal of the Korean Statistical Society
• /
• 제6권1호
• /
• pp.65-76
• /
• 1977

The relative importance of single-shot kill probabilities, rates of fire, weapon allocation strategies, and the size of initial force in warfare between two force with heterogeneous multiple weapon systems are considered by examining their effect on a natural measure of effectiveness, the expected number of survivors. Attrition equations are derived via stochastic formulation to represent the mean course of battle having an underlying probability distribution. It is assumed that each side uses indirect area fires. Level of intelligence activities are reflected in the availability of spontaneous information on the current enemy status. Depending on the availability of the information on the current enemy status, each participatory unit may follow 1) a prescribed attack pattern (fraction of the available units assigned to various enemy targets) or 2) an adaptive attack pattern depending on the enemy status at that time. Conditions for possible stalemate are discussed.

• 한국국방경영분석학회지
• /
• 제17권2호
• /
• pp.44-53
• /
• 1991

This paper is aimed to develop a Lanchester type combat model for the direct-fire engagement. This model incorporates number of combatants, inter-firing time, detection time by movement, detection probability by the signature of fire, where the inter-firing time and the detection time are assumed to follow a negative exponential distribution. The approach to modeling is as follows : in the process of an engagement, a combatant takes one of the states('observing' state or 'firing' state), a combatant is initially in the observing state, if the combatant detects a target, he changes his state from 'observing' to 'firing' and will cause attrition to the opposing forces. Thus this transition mechanism is embodied into the differential equation form with each transition rate. A limited examination of the validity has been conducted by comparison with the Monte-Carlo simulation model 'BAGSIM', and with a traditional Deterministic Lanchester model.

• 한국주거학회논문집
• /
• 제23권1호
• /
• pp.55-66
• /
• 2012

Present study investigates on the impact of resources and characteristics of the tenure choice for divorced women in Korea. The authors utilize the micro data from the Korea Census (2% sample) provided by the National Statistical Office. The authors apply the bivariate probit model to eliminate selection bias that could incur due to sample selectivity, from a chain of marital disruption and tenure choices. This study starts with a descriptive explanation of homeownership after divorce from 1985 to 2005. It concluded that divorce results in a substantial attrition of homeownership. The authors found that out for many women, divorce initiates a process of downward mobility on the housing ladder. The probability to own housing is much lower for divorced women than for women who are not divorced. The present study concludes by suggesting some policy implications for divorced women who have limited access to housing stability. The authors also suggest some future studies that can compensate the empirical limitations of the present study.

• 한국빅데이터학회지
• /
• 제8권1호
• /
• pp.23-34
• /
• 2023

대리운전 시장의 요금체계는 이용자의 대리운전 서비스 요청 시, 주위 대리운전 기사의 수, 대리운전 서비스 이용자 수, 날씨 등의 다양한 영향에 의해 실시간으로 변하는 Dynamic Pricing의 특징이 있다. 불확실한 변동성은 대리운전 서비스 요금을 상승시켜 고객의 이탈과 대리운전 기사의 배차거부를 유발하는 주된 원인이 되며, 이러한 문제를 해결하기 위해 적절한 수요를 예측하고 선제적으로 대응하기 위한 수요 예측모형의 도입이 요구된다. 본 연구에서는 대리운전 서비스 이력 데이터를 활용하여 지역별, 시간대별 대리운전 서비스 수요를 예측하는 모형을 제시한다. 이후 실제 대리운전 서비스 이력 데이터를 활용하여 시간과 요일에 따른 조건부 확률을 구축하고 이를 예측 모형과 결합한 Time-Series with Conditional Probability 방법론을 제안하였으며 실험을 통해 SARIMA, Prophet의 기존 시계열 모형보다 성능이 우수함을 검증하였다. 본 연구는 제안된 방법론을 통해 구축된 수요 예측 모형을 활용하여 대리운전 서비스의 단기 전략 수립에 활용할 수 있다는 시사점이 있다.