Non-informative Samples Research Articles

Superpopulation model inference for non-probability samples under informative sampling

Non-probability samples have been extensively applied in many various fields in recent years. However, it is difficult to infer the population from non-probability samples since selection probabilities of non-probability samples is unknown. Superpopulation modeling has been recently explored to make inference from non-probability samples. However, the existing superpopulation model approaches rely mostly on the noninformative sampling assumption. When sampling is informative, the sample distribution differs from the population distribution. Ignoring this point may result in biased estimation. In this paper, taking into account the informative sampling scheme, a superpopulation model approach for non-probability samples is proposed. Exponential model, logistic model and probit model are established to explain the informative sampling mechanism, respectively. The sample likelihood is derived to estimate the superpopulation model parameters for non-probability samples under various informative sampling models. The population mean estimator can be obtained based on the fitted superpopulation model from non-probability samples under different informative sampling approaches. The theoretical properties of the proposed estimator are established. Simulation results illustrate the performance of our proposed method for different informative sampling models and sample sizes. Also, the proposed method is applied to the data from the National Health and Nutrition Examination Survey.

Optimal Predictors of General Small Area Parameters Under an Informative Sample Design Using Parametric Sample Distribution Models

Abstract Two challenges in small area estimation occur when (i) the sample selection mechanism depends on the outcome variable and (ii) the parameter of interest is a nonlinear function of the response variable in the assumed model. If, given the values of the model covariates, the sample selection mechanism depends on the model response variable, the design is said to be informative for the model. Pfeffermann and Sverchkov (2007) develop a small area estimation procedure for informative sampling, focusing on the prediction of small area means. Molina and Rao (2010) develop a small area estimation procedure for general parameters that are nonlinear functions of the model response variable. The method of Molina and Rao assumes noninformative sampling. We combine these two approaches to develop a procedure for the estimation of general parameters in small areas under informative sampling. We introduce a parametric bootstrap MSE estimator that is appropriate for an informative sample design. We evaluate the validity of the proposed procedures through extensive simulation studies and illustrate the procedures utilizing Mexico’s income data.

Sequential adaptive strategies for sampling rare clustered populations

A new class of sampling strategies is proposed that can be applied to population-based surveys targeting a rare trait that is unevenly spread over an area of interest. Our proposal is characterised by the ability to tailor the data collection to specific features and challenges of the survey at hand. It is based on integrating an adaptive component into a sequential selection, which aims both to intensify the detection of positive cases, upon exploiting the spatial clustering, and to provide a flexible framework to manage logistics and budget constraints. A class of estimators is also proposed to account for the selection bias, that are proved unbiased for the population mean (prevalence) as well as consistent and asymptotically Normal distributed. Unbiased variance estimation is also provided. A ready-to-implement weighting system is developed for estimation purposes. Two special strategies included in the proposed class are presented, that are based on the Poisson sampling and proved more efficient. The selection of primary sampling units is also illustrated for tuberculosis prevalence surveys, which are recommended in many countries and supported by the World Health Organisation as an emblematic example of the need for an improved sampling design. Simulation results are given in the tuberculosis application to illustrate the strengths and weaknesses of the proposed sequential adaptive sampling strategies with respect to traditional cross-sectional non-informative sampling as currently suggested by World Health Organisation guidelines.

Unconditional empirical likelihood approach for analytic use of public survey data

AbstractModeling survey data often requires having the knowledge of design and weighting variables. With public‐use survey data, some of these variables may not be available for confidentiality reasons. The proposed approach can be used in this situation, as long as calibrated weights and variables specifying the strata and primary sampling units are available. It gives consistent point estimation and a pivotal statistics for testing and confidence intervals. The proposed approach does not rely on with‐replacement sampling, single‐stage, negligible sampling fractions, or noninformative sampling. Adjustments based on design effects, eigenvalues, joint‐inclusion probabilities or bootstrap, are not needed. The inclusion probabilities and auxiliary variables do not have to be known. Multistage designs with unequal selection of primary sampling units are considered. Nonresponse can be easily accommodated if the calibrated weights include reweighting adjustment for nonresponse. We use an unconditional approach, where the variables and sample are random variables. The design can be informative.

Remaining Useful Life Prediction of Aircraft Engines with Variable Length Input Sequences

Remaining useful life (RUL) is the expected remaining operating life of an asset until it can no longer perform its intended function. The 2021 PHM Society Data Challenge posed the problem of estimating the RUL of aircraft engines with various competing failure modes and underlying degradation trajectories. In this work, we describe the approach and solution to the challenge where we map from the multivariate time series sensor readings to the remaining useful life, measured in the remaining number of flight cycles until failure. The proposed solution utilizes a deep convolutional neural network that can take inputs of variable length. Furthermore, we preprocess the data according to a normalization procedure that help reveal the degradation trend that is obfuscated by the continuously varying flight conditions. The normalization procedure involves training a feedforward neural network on a non-degraded subset of the data that maps from the flight conditions to the sensor outputs. The difference between the expected sensor readings and the actual observations is then interpreted as the extent of deviation from normal, i.e., degradation. Finally, we describe the sampling techniques which is designed to reduce the number of non-informative samples fed to the neural network.

ASNet: Auto-Augmented Siamese Neural Network for Action Recognition.

Human action recognition methods in videos based on deep convolutional neural networks usually use random cropping or its variants for data augmentation. However, this traditional data augmentation approach may generate many non-informative samples (video patches covering only a small part of the foreground or only the background) that are not related to a specific action. These samples can be regarded as noisy samples with incorrect labels, which reduces the overall action recognition performance. In this paper, we attempt to mitigate the impact of noisy samples by proposing an Auto-augmented Siamese Neural Network (ASNet). In this framework, we propose backpropagating salient patches and randomly cropped samples in the same iteration to perform gradient compensation to alleviate the adverse gradient effects of non-informative samples. Salient patches refer to the samples containing critical information for human action recognition. The generation of salient patches is formulated as a Markov decision process, and a reinforcement learning agent called SPA (Salient Patch Agent) is introduced to extract patches in a weakly supervised manner without extra labels. Extensive experiments were conducted on two well-known datasets UCF-101 and HMDB-51 to verify the effectiveness of the proposed SPA and ASNet.

Re-SSS: Rebalancing Imbalanced Data Using Safe Sample Screening

Different samples can have different effects on learning support vector machine (SVM) classifiers. To rebalance an imbalanced dataset, it is reasonable to reduce non-informative samples and add informative samples for learning classifiers. Safe sample screening can identify a part of non-informative samples and retain informative samples. This study developed a resampling algorithm for Rebalancing imbalanced data using Safe Sample Screening (Re-SSS), which is composed of selecting Informative Samples (Re-SSS-IS) and rebalancing via a Weighted SMOTE (Re-SSS-WSMOTE). The Re-SSS-IS selects informative samples from the majority class, and determines a suitable regularization parameter for SVM, while the Re-SSS-WSMOTE generates informative minority samples. Both Re-SSS-IS and Re-SSS-WSMOTE are based on safe sampling screening. The experimental results show that Re-SSS can effectively improve the classification performance of imbalanced classification problems.

Rescue of Non-Informative Circulating Tumor DNA to Monitor the Mutational Landscape in NSCLC

In non-small cell lung cancer (NSCLC) the usage of plasma-derived circulating tumor DNA (ctDNA) have come into focus to obtain a comprehensive genetic profile of a given lung cancer. Despite the usage of specific sampling tubes, archived plasma samples as well as inappropriately treated blood samples still cause a loss of information due to cell lysis and contamination with cellular DNA. Our aim was to establish a reliable protocol to rescue ctDNA from such non-informative samples to monitor the mutational landscape in NSCLC. As a proof-of-concept study we used archived plasma samples derived from whole blood EDTA samples of 51 patients suffering from NSCLC. Analysis of the isolated plasma DNA determined only a small fraction of ctDNA in a range of 90–250 bp. By applying a specific purification procedure, we were able to increase the informative ctDNA content and improve in a cohort of 42 patients the detection of driver mutations from 32% to 79% of the mutations found in tissue biopsies. Thus, we present here an easy to perform, time and cost effective procedure to rescue non-informative ctDNA samples, which is sufficient to detect oncogenic mutations in NGS approaches and is therefore a valuable technical improvement for laboratories handling liquid biopsy samples.

Parametric prediction of finite population total under Informative sampling and nonignorable nonresponse

Abstract In this paper, we combine two methodologies used in the model-based survey sampling, namely the prediction of the finite population total, named T, under informative sampling and full response, see Sverchkov and Pfeffermann (2004), and the prediction of T with a noninformative sampling design and the nonignorable nonresponse mechanism, see Eideh (2012). The former approach involves the dependence of the first order inclusion probabilities on the study variable, while the latter involves the dependence of the probability of nonresponse on unobserved or missing observations. The main aim of the paper is to consider how to account for the joint effects of informative sampling designs and not-missing-at-random response mechanism in statistical models for complex survey data. For this purpose, theoretically, we use the response distribution and relationships between the moments of the superpopoulation, the sample, sample-complement, response, and nonresponse distributions for the prediction of finite population totals, see Eideh (2016). The derived parametric predictors of T use the observation for the response set of the study variable or variable of interest, values of auxiliary variables and their population totals, sampling weights, and propensity scores. An interesting outcome of the T study is that most predictors known from model-based survey sampling can be derived as a special case from this general theory, see Chambers and Clark (2012).

The Pseudo Maximum Likelihood Estimator for Quantiles of Survey Variables

Abstract The pseudo maximum likelihood approach has been employed in multilevel models for analysis of complex survey data. This approach may be inappropriate for many survey variables that are nonsymmetrically distributed with skewness and multimodality characteristics. This article intends to fill this gap in the literature by developing a pseudolikelihood estimator for quantiles of survey variables in the quantile regression framework. This approach is illustrated using a Monte Carlo simulation study and the body mass index data from the 1998–1999 Early Childhood Longitudinal Study. Results show that the proposed estimator is consistent and approximately unbiased for both informative and noninformative sampling designs.

Variance reduction using a non-informative sampling design

Variance reduction using a non-informative sampling design

Small area estimation of general parameters under complex sampling designs

When the probabilities of selecting individuals (units) for the sample depend on the outcome values, the selection mechanism is said to be informative. Under informative selection, individuals with certain outcome values appear more often in the sample and, as a consequence, usual inference based on the actual sample without appropriate weighting might be strongly biased. An extension of the empirical best (EB) method for estimation of general non-linear parameters in small areas that handles informative selection by incorporating the sampling weights is proposed. Properties of this new method under complex sampling designs, including informative selection, are analyzed. Results confirm that the proposed weighted estimators significantly reduce the bias of unweighted EB estimators under informative sampling, and compare favorably under non-informative sampling. The proposed method is illustrated through an application to poverty mapping in a State from Mexico.

Cross-Platform Emerging Topic Detection and Elaboration from Multimedia Streams

With the explosive growth of online media platforms in recent years, it becomes more and more attractive to provide users a solution of emerging topic detection and elaboration. And this posts a real challenge to both industrial and academic researchers because of the overwhelming information available in multiple modalities and with large outlier noises. This article provides a method on emerging topic detection and elaboration using multimedia streams cross different online platforms. Specifically, Twitter, New York Times and Flickr are selected for the work to represent the microblog, news portal and imaging sharing platforms. The emerging keywords of Twitter are firstly extracted using aging theory. Then, to overcome the nature of short length message in microblog, Robust Cross-Platform Multimedia Co-Clustering (RCPMM-CC) is proposed to detect emerging topics with three novelties: 1) The data from different media platforms are in multimodalities; 2) The coclustering is processed based on a pairwise correlated structure, in which the involved three media platforms are pairwise dependent; 3) The noninformative samples are automatically pruned away at the same time of coclustering. In the last step of cross-platform elaboration, we enrich each emerging topic with the samples from New York Times and Flickr by computing the implicit links between social topics and samples from selected news and Flickr image clusters, which are obtained by RCPMM-CC. Qualitative and quantitative evaluation results demonstrate the effectiveness of our method.

A joint model for multistate disease processes and random informative observation times, with applications to electronic medical records data.

Multistate models are used to characterize individuals' natural histories through diseases with discrete states. Observational data resources based on electronic medical records pose new opportunities for studying such diseases. However, these data consist of observations of the process at discrete sampling times, which may either be pre-scheduled and non-informative, or symptom-driven and informative about an individual's underlying disease status. We have developed a novel joint observation and disease transition model for this setting. The disease process is modeled according to a latent continuous-time Markov chain; and the observation process, according to a Markov-modulated Poisson process with observation rates that depend on the individual's underlying disease status. The disease process is observed at a combination of informative and non-informative sampling times, with possible misclassification error. We demonstrate that the model is computationally tractable and devise an expectation-maximization algorithm for parameter estimation. Using simulated data, we show how estimates from our joint observation and disease transition model lead to less biased and more precise estimates of the disease rate parameters. We apply the model to a study of secondary breast cancer events, utilizing mammography and biopsy records from a sample of women with a history of primary breast cancer.

Incremental Import Vector Machines for Classifying Hyperspectral Data

In this paper, we propose an incremental learning strategy for import vector machines (IVM), which is a sparse kernel logistic regression approach. We use the procedure for the concept of self-training for sequential classification of hyperspectral data. The strategy comprises the inclusion of new training samples to increase the classification accuracy and the deletion of noninformative samples to be memory and runtime efficient. Moreover, we update the parameters in the incremental IVM model without retraining from scratch. Therefore, the incremental classifier is able to deal with large data sets. The performance of the IVM in comparison to support vector machines (SVM) is evaluated in terms of accuracy, and experiments are conducted to assess the potential of the probabilistic outputs of the IVM. Experimental results demonstrate that the IVM and SVM perform similar in terms of classification accuracy. However, the number of import vectors is significantly lower when compared to the number of support vectors, and thus, the computation time during classification can be decreased. Moreover, the probabilities provided by IVM are more reliable, when compared to the probabilistic information, derived from an SVM's output. In addition, the proposed self-training strategy can increase the classification accuracy. Overall, the IVM and its incremental version is worthwhile for the classification of hyperspectral data.

An active learning based classification strategy for the minority class problem: application to histopathology annotation

BackgroundSupervised classifiers for digital pathology can improve the ability of physicians to detect and diagnose diseases such as cancer. Generating training data for classifiers is problematic, since only domain experts (e.g. pathologists) can correctly label ground truth data. Additionally, digital pathology datasets suffer from the "minority class problem", an issue where the number of exemplars from the non-target class outnumber target class exemplars which can bias the classifier and reduce accuracy. In this paper, we develop a training strategy combining active learning (AL) with class-balancing. AL identifies unlabeled samples that are "informative" (i.e. likely to increase classifier performance) for annotation, avoiding non-informative samples. This yields high accuracy with a smaller training set size compared with random learning (RL). Previous AL methods have not explicitly accounted for the minority class problem in biomedical images. Pre-specifying a target class ratio mitigates the problem of training bias. Finally, we develop a mathematical model to predict the number of annotations (cost) required to achieve balanced training classes. In addition to predicting training cost, the model reveals the theoretical properties of AL in the context of the minority class problem.ResultsUsing this class-balanced AL training strategy (CBAL), we build a classifier to distinguish cancer from non-cancer regions on digitized prostate histopathology. Our dataset consists of 12,000 image regions sampled from 100 biopsies (58 prostate cancer patients). We compare CBAL against: (1) unbalanced AL (UBAL), which uses AL but ignores class ratio; (2) class-balanced RL (CBRL), which uses RL with a specific class ratio; and (3) unbalanced RL (UBRL). The CBAL-trained classifier yields 2% greater accuracy and 3% higher area under the receiver operating characteristic curve (AUC) than alternatively-trained classifiers. Our cost model accurately predicts the number of annotations necessary to obtain balanced classes. The accuracy of our prediction is verified by empirically-observed costs. Finally, we find that over-sampling the minority class yields a marginal improvement in classifier accuracy but the improved performance comes at the expense of greater annotation cost.ConclusionsWe have combined AL with class balancing to yield a general training strategy applicable to most supervised classification problems where the dataset is expensive to obtain and which suffers from the minority class problem. An intelligent training strategy is a critical component of supervised classification, but the integration of AL and intelligent choice of class ratios, as well as the application of a general cost model, will help researchers to plan the training process more quickly and effectively.

Linear estimation and prediction under model-design approach with small area effects

We consider a problem of small area estimation under a mixed linear model with area-specific auxiliary variables and random area effects. The model design approach is consistently pursued. The mathematical results of this paper support the view (not generally accepted in the literature) that while using a super-population model, even under non-informative sampling, the sample selection process cannot be completely ignored. Estimators and predictors under consideration are linear in observations, that is in variables Z i ’s, where Z i is the value of the variable of interest if i is in the sample and zero otherwise. This notion of linearity is different from that prevailing in the literature. Unbiasedness and optimality are understood with respect to both the model and the sampling design. We consider general sampling designs, for which sample sizes in small areas can be random. We show that the best linear unbiased estimators (BLUEs) and best linear unbiased predictors (BLUPs) in general do not exist. However, they do exist if the sample sizes in small areas are fixed. Moreover, we prove that such designs are optimal. Empirical versions (EBLUE and EBLUPs) are also derived and numerically tested. In simulation experiments, we examine the mean square error of estimates/predictors, the coverage rates of confidence intervals and the predictive power of auxiliary variables. Rather unexpectedly, the proposed predictors turn out to be quite robust against model misspecification. A special case of Bernoulli sampling is examined in detail as an illustrative example.

Allele-specific expression of TGFBR1 in colon cancer patients

The genetic component of colorectal cancer (CRC) predisposition has been only partially explained. We recently suggested that a subtle decrease in the expression of one allele of the TGFBR1 gene was a heritable quantitative trait predisposing to CRC. Here, we refined the measurements of allele-specific expression (ASE) of TGFBR1 in a population-based series of CRC patients and controls. Five single-nucleotide polymorphisms (SNPs) in the 3′-untranslated region of the gene were genotyped and used for ASE determination by pyrosequencing. After eliminating non-informative samples and samples with RNA of insufficient quality 109 cases and 125 controls were studied. Allelic ratios ranged between 0.74 and 1.69 without evidence of bimodality or cutoff points for ‘ASE’ versus ‘non-ASE’. Treating ASE as a continuous variable, cases had non-significantly different values than controls (P = 0.081 when comparing means by permutation test). However, cases had significantly higher ASE values when comparing medians by permutation test (P = 0.0027) and when using Wilcoxon test (P = 0.0094). We conclude that with the present-day technology, ASE differences between individuals and between cases and controls are too subtle to be used to assess CRC risk. More advanced technology is expected to resolve this issue as well as the low informativity caused by the limited heterozygosity of transcribed SNPs.

Fitting time series models for longitudinal survey data under informative sampling

The purpose of this paper is to account for informative sampling in fitting time series models, and in particular an autoregressive model of order one, for longitudinal survey data. The idea behind the proposed approach is to extract the model holding for the sample data as a function of the model in the population and the first-order inclusion probabilities, and then fit the sample model using maximum-likelihood, pseudo-maximum-likelihood and estimating equations methods. A new test for sampling ignorability is proposed based on the Kullback–Leibler information measure. Also, we investigate the issue of the sensitivity of the sample model to incorrect specification of the conditional expectations of the sample inclusion probabilities. The simulation study carried out shows that the sample-likelihood-based method produces better estimators than the pseudo-maximum-likelihood method, and that sensitivity to departures from the assumed model is low. Also, we find that both the conventional t-statistic and the Kullback–Leibler information statistic for testing of sampling ignorability perform well under both informative and noninformative sampling designs.

A technical note for using microsatellite DNA analyses in haploid male DNA pools of social Hymenoptera

The analyses of haploid male DNA pooled (MDP) samples can increase the cost efficiency in population genetic studies for a variety of applications in fundamental and applied screening of social Hymenoptera populations. Depending on the research problem addressed it can reduce the number of genotyped samples per study by orders of magnitude. In this study we show the feasibility of MDPs for assessing population allele frequencies, mother queen genotypes and mutation rates for microsatellite loci in the honeybee (Apis mellifera). The main purpose of MDP genotyping is not to replace individual based analyses but to discard non-informative samples and monomorphic loci from further analyses.