Some results on estimating a change-point using non-parametric type statistics

A Mann-Whitney type statistic is used to estimate a change-point when a change, at an unknown point in a sequence of random variables, has taken place. This estimate is compared, using Monte Carlo techniques, with the normal theory maximum likelihood estimate, when a location change has occurred, for different underlying distributions ranging from the normal to the long tailed “normal over uniform” distribution. The distribution of the Mann-Whitney type estimate remains fairly constant over the various distributions. Two generalisations of the statistic are considered and investigated.     

Modified mantel-haenszel procedures for matched pairs

In the prospective study of a finely stratified population, one individual from each stratum is chosen at random for the “treatment” group and one for the “non-treatment” group. For each individual the probability of failure is a logistic function of parameters designating the stratum, the treatment and a covariate. Uniformly most powerful unbiased tests for the treatment effect are given. These tests are generally cumbersome but, if the covariate is dichotomous, the tests and confidence intervals are simple. Readily usable (but non-optimal) tests are also proposed for poly-tomous covariates and factorial designs. These are then adapted to retrospective studies (in which one “success” and one “failure” per stratum are sampled). Tests for retrospective studies with a continuous “treatment” score are also proposed.     

EXACT TWO-SAMPLE MEDIAN TESTS WHEN ONE "SAMPLE" VALUE POSSIBLY FROM A DIFFERENT POPULATION

Random samples are assumed for the univariate two-sample problem. Sometimes this assumption may be violated in that an observation in one “sample”, of size m, is from a population different from that yielding the remaining m—1 observations (which are a random sample). Then, the interest is in whether this random sample of size m—1 is from the same population as the other random sample. If such a violation occurs and can be recognized, and also the non-conforming observation can be identified (without imposing conditional effects), then that observation could be removed and a two-sample test applied to the remaining samples. Unfortunately, satisfactory procedures for such a removal do not seem to exist. An alternative approach is to use two-sample tests whose significance levels remain the same when a non-conforming observation occurs, and is removed, as for the case where the samples were both truly random. The equal-tail median test is shown to have this property when the two “samples” are of the same size (and ties do not occur).     

Asymptotic study of the change-point mle in multivariate Gaussian families under contiguous alternatives

The problem of estimating an unknown change-point in the mean vector or covariance matrix of a sequence of independent multivariate Gaussian random variables is considered. Adapting the estimation methodology that Hinkley pursued for the case of abrupt changes, we develop theory for deriving the asymptotic distribution of the maximum likelihood estimator of the change-point when the amount of change is a function of the sample size and goes to zero in a smooth fashion as the sample size goes to infinity, yielding a contiguous change-point model. Simulations have been performed to illustrate the closeness of the asymptotic distribution with the empirical distribution, and to evaluate its robustness to departures from normality for reasonable sample sizes as well as parameter changes. Finally, we apply the methodology to estimate the change-point in the daily log-returns data of BLS (BellSouth) and VZ (Verizon) from NYSE.     

A self-normalization test for a change-point in the shape parameter of a gamma distributed sequence

In this paper, a parametric change-point problem encountered in analyzing a gamma distributed sequence is considered. We propose a self-normalization based CUSUM type test statistic to detect the presence of a change-point, and obtain its limiting null distribution. The CUSUM-based procedure for detecting the location of this change-point is also given. At the same time, simulation results are provided, which show that our procedures are effective.     

Inference for post-change mean by a CUSUM procedure

In this paper, the inference problem for the post-change mean is considered after a change is detected by a CUSUM process in a sequence of independent normal variables. The change-point is estimated as the maximum likelihood estimate at the reference value and the post-change mean is estimated as the sample mean after the change-point estimate. By assuming the change-point is large and the monitoring limit approaches infinity, the first-order bias of the post-change mean estimate and a corrected asymptotic normal pivot are derived conditioning on that a change is detected. Local approximations for small reference value and post-change mean are obtained for numerical evaluation.     

On robustness of the Shiryaev–Roberts change-point detection procedure under parameter misspecification in the post-change distribution

The gist of the quickest change-point detection problem is to detect the presence of a change in the statistical behavior of a series of sequentially made observations, and do so in an optimal detection-speed-versus-“false-positive”-risk manner. When optimality is understood either in the generalized Bayesian sense or as defined in Shiryaev's multi-cyclic setup, the so-called Shiryaev–Roberts (SR) detection procedure is known to be the “best one can do”, provided, however, that the observations’ pre- and post-change distributions are both fully specified. We consider a more realistic setup, viz. one where the post-change distribution is assumed known only up to a parameter, so that the latter may be misspecified. The question of interest is the sensitivity (or robustness) of the otherwise “best” SR procedure with respect to a possible misspecification of the post-change distribution parameter. To answer this question, we provide a case study where, in a specific Gaussian scenario, we allow the SR procedure to be “out of tune” in the way of the post-change distribution parameter, and numerically assess the effect of the “mistuning” on Shiryaev's (multi-cyclic) Stationary Average Detection Delay delivered by the SR procedure. The comprehensive quantitative robustness characterization of the SR procedure obtained in the study can be used to develop the respective theory as well as to provide a rational for practical design of the SR procedure. The overall qualitative conclusion of the study is an expected one: the SR procedure is less (more) robust for less (more) contrast changes and for lower (higher) levels of the false alarm risk.     

Maximum likelihood estimator in a multi-phase random regression model

We consider a random regression model with several-fold change-points. The results for one change-point are generalized. The maximum likelihood estimator of the parameters is shown to be consistent, and the asymptotic distribution for the estimators of the coefficients is shown to be Gaussian. The estimators of the change-points converge, with n ^?1 rate, to the vector whose components are the left end points of the maximizing interval with respect to each change-point. The likelihood process is asymptotically equivalent to the sum of independent compound Poisson processes.     

AN INVENTORY PROBLEM APPLIED TO A RENTAL SITUATION

An inventory problem, applied to a rental situation business, has been considered. If no item is in stock when a demand occurs, the company borrows the units from other concerns in the same line of business. The profit function has been calculated and it has been shown how the inventory level increases with penalty cost. A review of the literature on inventory control reveals that not much work has been done in holding inventories of rental items. In a recent paper, Tainiter (1964) considered the situation in which a company rents out items such as cars, trucks, farm equipment, books, furniture, etc., and obtained the profit function by taking “rental-out” time as a negative exponential and demand as a general random variable. The model is equally applicable to companies which rent out “service personnel”, repair men, taxi cabs, etc. “We consider a company renting out items to customers. The company starts its business by purchasing a total number of M items in the inventory. The term inventory, defined by Arrow, Earlin and Scarf (1958) as the stock of goods which is kept for future sale or production, is applicable here. Whenever a demand occurs the item is rented out immediately, if it is available in the stock. But if the inventory is zero, i.e. all the items are rented out, the demand will be satisfied by borrowing items from other companies which are dealing in the same line of business. For example, a manufacturer of refrigerators maintains and repairs his product at the customer's house after sale. If a complaint arrives when no repairmen are available, the company will “borrow” repairmen from elsewhere and will attend to the complaint immediately. The borrowing cost may be negative or positive, representing a penalty or a profit. On the other hand if the company does not borrow and the customer has to wait (and such situations occur very often) the loss of the customer's goodwill may occur. It is also not possible to keep large numbers of items because of the storage costs and tied up capital. The problem is then to devise an optimal policy such that the profits of the company are maximized.     

Bayesian change-point problem using Bayes factor with hierarchical prior distribution

We consider the hierarchical Bayesian models of change-point problem in a sequence of random variables having either normal population or skew-normal population. Further, we consider the problem of detecting an influential point concerning change point using Bayes factors. Our proposed models are illustrated with the real data example, the annual flow volume data of Nile River at Aswan from 1871 to 1970. The result using our proposed models indicated the largest influential observation in the year 1888 among outliers. We have shown that it is useful to measure the influence of observations on Bayes factors. Here, we consider omitting single observation as well.     

Admissibility of generalized bayes and pitman estimates in the non-regular family

In this paper we have considered the problem of finding admissible estimates for a fairly general class of parametric functions in the so called “non-regular” type of densities Following Karlin s (1958) technique, we have established the ad-missibility of generalized Bayes estimates and Pitman estimates. Some examples are discussed.     

The practice of mining gegstatistics in 1980

The scope of the term “Mining Geostatistics” is defined in the context of a brief historical review of the topic. Assumptions andtools adapted from probability theory are applied first to linear estimation problems leading to the iging” estimator and then shown capable of extension to non- ationary and non-linear situa­tions, in which conditional simulation, conditional distributions and “disjunctive Kriging” become applicable. Directions for further research work are indicated regarding variogram identification and modelling of random functions where stationarity or linearity may not hold. The brief review suggests that the mutual benefits of practice and theory will continue to develop.     

Multiple change-point estimation with U-statistics

We consider a multiple change-point problem: a finite sequence of independent random variables consists of segments given by a known number of the so-called change-points such that the underlying distribution differs from segment to segment. The task is to estimate these change-points under no further assumptions on the within-segment distributions. In this completely nonparametric framework the proposed estimator is defined as the maximizing point of weighted multivariate U-statistic processes. Under mild moment conditions we prove almost sure convergence and the rate of convergence.     

SUPRA-BAYESIAN POOLING OF PRIORS LINKED BY A DETERMINISTIC SIMULATION MODEL

Deterministic simulation models are used to guide decision-making and enhance understanding of complex systems such as disease transmission, population dynamics, and tree plantation growth. Bayesian inference about parameters in deterministic simulation models can require the pooling of expert opinion. One class of approaches to pooling expert opinion in this context is supra-Bayesian pooling, in which expert opinion is treated as data for an ultimate decision maker. This article details and compares two supra-Bayesian approaches—“event updating” and “parameter updating.” The suitability of each approach in the context of deterministic simulation models is assessed based on theoretical properties, performance on examples, and the selection and sensitivity of required hyperparameters. In general, we favor a parameter updating approach because it uses more intuitive hyperparameters, it performs sensibly on examples, and because the alternative event updating approach fails to exhibit a desirable property (relative propensity consistency) in all cases. Inference in deterministic simulation models is an increasingly important statistical and practical problem, and supra-Bayesian methods represent one viable option for achieving a sensible pooling of expert opinion.     

Change-point mle in the rate of exponential sequences with application to Indonesian seismological data

In this paper, we derive explicit computable expressions for the asymptotic distribution of the maximum likelihood estimate of an unknown change-point in a sequence of independently and exponentially distributed random variables. First we state and prove a theorem that shows asymptotic equivalence of the change-point mle for the cases of both known and unknown parameters, respectively. Thereafter, the computational form of the asymptotic distribution of the change-point mle is derived for the case of known parameter situation only. Simulations show that the distribution for the known case applies very well to the case where the parameters are estimated. Further, it is seen from simulations that the derived unconditional mle shows better performance compared to the conditional solution of Cobb. Application of change detection methodology and the derived estimation methodology show strong support in favor the dynamic triggering hypothesis for seismic faults in Sumatra, Indonesia region.     

Tests of statistical hypotheses and their use in studies of natural phenomena

Contrary to ideas suggested by the title of the conference at which the present paper was presented, the author is not aware of a conceptual difference between a “test of a statistical hypothesis” and a “test of significance” and uses these terms interchangeably. A study of any serious substantive problem involves a sequence of incidents at which one is forced to pause and consider what to do next. In an effort to reduce the frequency of misdirected activities one uses statistical tests. The procedure is illustrated on two examples: (i) Le Cam’s (and associates’) study of immunotherapy of cancer and (ii) a socio-economic experiment relating to low-income homeownership problems.     

Change Detection in INAR(p) Processes Against Various Alternative Hypotheses

Change in the coefficients or the mean of the innovation of an INAR(p) process is a sign of disturbance that is important to detect. The proposed methods can test for change in any one of these quantities separately, or in any collection of them. They make both one-sided and two-sided tests possible, furthermore, they can be used to test against the “epidemic” alternative. The tests are based on a CUSUM process using CLS estimators of the parameters. Under the one-sided and two-sided alternatives, consistency of the tests is proved and the properties of the change-point estimator are also explored.     

Quota fulfillment times

An example of the classical occupancy problem is to sample with replacement from an urn containing several colours of balls and count the number of balls sampled until a given number of “quotas” are filled. This and the corresponding random variable for sampling without replacement will be referred to as quota fulfillment times. Asymptotic and exact methods for computing moments and distributions are given in this paper. Moments of quota fulfillment times are related to moments of order statistics of beta and gamma random variables. Most of the results for sampling without replacement and some of the results for sampling with replacement are believed to be new. Some other known sampling-with-replacement results are given for comparative purposes.     

Modelling of dependence between critical failure and preventive maintenance: The repair alert model

We consider the competing risks problem for a repairable unit which at each sojourn may be subject to either a critical failure, or a preventive maintenance (PM) action, where the latter will prevent the failure. It is reasonable to expect a dependence between the failure mechanism and the PM regime. The paper presents a new model, called the repair alert model, for handling such cases. This model is a special case of random signs censoring, which was introduced by Roger Cooke [1993. The total time on test statistic and age-dependent censoring. Statist. Probab. Lett., 18, 307–312]. The pleasant feature of random signs censoring is that the marginal distribution of the failure time is identifiable. The repair alert model introduces the so-called repair alert function, which characterizes the “alertness” of the maintenance crew, and which is shown to be uniquely identifiable from field data. Statistical estimation is considered both nonparametrically and parametrically.     

Convolution of fuzzy distributions in decision-making

This paper deals with a decision-making problem that has hardly been addressed so far. Frequently, management is challenged to deal with sums of random variables, the distribution of which is not fully known. For example, a high return on an investment project may be “more likely” than a low one in the first period but “less likely” in the second period; however, the investment decision has to be based on the sum of the two retums. The paper contains some essential theorems concerning the convolution of distribution functions for the case of fuzzy random variables. Moreover, an example is given illustrating the incorporation of new Linear Partial Information (LPI) and the transition from an a priori to a posteriori convolution of fuzzy distribution functions.