Journal of Data Science

Pupillary response behavior (PRB) refers to changes in pupil diameter in response to simple or complex stimuli. There are underlying, unique patterns hidden within complex, high-frequency PRB data that can be utilized to classify visual impairment, but those patterns cannot be described by traditional summary statistics. For those complex high-frequency data, Hurst exponent, as a measure of long-term memory of time series, becomes a powerful tool to detect the muted or irregular change patterns. In this paper, we proposed robust estimators of Hurst exponent based on non-decimated wavelet transforms. The properties of the proposed estimators were studied both theoretically and numerically. We applied our methods to PRB data to extract the Hurst exponent and then used it as a predictor to classify individuals with different degrees of visual impairment. Compared with other standard wavelet-based methods, our methods reduce the variance of the estimators and increase the classification accuracy.

Researchers and practitioners of many areas of knowledge frequently struggle with missing data. Missing data is a problem because almost all standard statistical methods assume that the information is complete. Consequently, missing value imputation offers a solution to this problem. The main contribution of this paper lies on the development of a random forest-based imputation method (TI-FS) that can handle any type of data, including high-dimensional data with nonlinear complex interactions. The premise behind the proposed scheme is that a variable can be imputed considering only those variables that are related to it using feature selection. This work compares the performance of the proposed scheme with other two imputation methods commonly used in literature: KNN and missForest. The results suggest that the proposed method can be useful in complex scenarios with categorical variables and a high volume of missing values, while reducing the amount of variables used and their corresponding preliminary imputations.

To date, many gene set analysis (GSA) approaches have been developed for identifying differentially expressed gene sets using microarray data. However, these methods are not directly

Unit root tests that are in common use today tend to over-reject the stationarity of economic ratios like the consumption-income ratio or rates like the average tax rate. The meaning of a unit root in such bounded series is not very clear. We use a mixed-frequency regression technique to develop a test for the null hypothesis that a series is stationary. The focus is on regression relationships, not so much on individual series. What is noteworthy about this moving average (MA) unit root test, denoted as z(MA) test, based on a variance-difference, is that, instead of having to deal with non-standard distributions, it takes testing back to normal distribution and offers a way to increase power without having to increase the sample size substantially. Monte Carlo simulations show minimal size distortions even when the AR root is close to unity and the test offers substantial gains in power relative to some popular tests against near-null alternatives in moderate size samples. Applying this test to log of consumption-income ratio of 21 OECD countries shows that the z(MA) test favors stationarity of 15 series, KPSS test 8 series, Johansen test 6 series and ADF test 5 series.

In this paper, we advance new families of bivariate copulas constructed by distributional distortions of existing bivariate copulas. The distortions under consideration are based on the unit gamma distribution of two forms. When the initial copula is Archimedean, the induced copula is also Archimedean under the admissible parameter space. Properties such as Kendall’s tau coefficient, tail dependence coefficients and tail orders for the new families of copulas are derived. An empirical application to economic indicator data is presented.

This paper proposes a new asymptotically valid stationary bootstrap procedure to obtain multivariate forecast densities in unrestricted vector autoregressive models. The proposed method is not based on either backward or forward representations, so it can be used for both Gaussian and non-Gaussian models. Also, it is computationally more efficient compared to the available resampling methods. The finite sample performance of the proposed method is illustrated by extensive Monte Carlo studies as well as a real-data example. Our records reveal that the proposed method is a good competitor or even better than the existing methods based on backward and/or forward representations.

Analysing seasonality in count time series is an essential application of statistics to predict phenomena in different fields like economics, agriculture, healthcare, environment, and climatic change. However, the information in the existing literature is scarce regarding the performances of relevant statistical models. This study provides the Yule-Walker (Y-W), Conditional Least Squares (CLS), and Maximum Likelihood Estimation (MLE) for First-order Non-negative Integer-valued Autoregressive, INAR(1), process with Poisson innovations with different monthly means. The performance of Y-W, CLS, and MLE are assessed by the Monte Carlo simulation method. The performance of this model is compared with another seasonal INAR(1) model by reproducing the monthly number of rainy days in the Blackwater River watershed located in coastal Virginia. Two forecast-coherent methods in terms of mode and probability function are applied to make predictions. The models’ performances are assessed using the Root Mean Square Error and Index of Agreement criteria. The results reveal the similar performance of Y-W, CLS, and MLE for estimating the parameters of data sets with larger sample size and values of α close to unite root. Moreover, the results indicate that INAR(1) with different monthly Poisson innovations is more appropriate for modelling and predicting seasonal count time series.

In this paper, parameter estimation for the power Lomax distribution is studied with different methods as maximum likelihood, maximum product spacing, ordinary least squares, weighted least squares, Cramér–von Mises and Bayesian estimation by Markov chain Monte Carlo (MCMC). Robust estimation of the stress-strength model for the Power Lomax distribution is discussed. We propose that the method of maximum product of spacing for reliable estimation of stress-strength model as an alternative method to maximum likelihood and Bayesian estimation methods. A numerical study using real data and Monte Carlo Simulation is performed to compare between different methods.

Analyzing time to event data arises in a number of fields such as Biology and Engineering. A common feature of this data is that, the exact failure time for all units may not be observable. Accordingly, several types of censoring were presented. Progressive censoring allows units to be randomly removed before the terminal point of the experiment. Marshall-Olkin bivariate lifetime distribution was first introduced in 1967 using the exponential distribution. Recently, bivariate Marshall-Olkin Kumaraswamy lifetime distribution was derived. This paper derives the likelihood function under progressive type-I censoring for the bivariate Marshall-Olkin family in general and applies it on the bivariate Kumaraswamy lifetime distribution. Maximum likelihood estimators of model parameters were derived. Simulation study and a real data set are presented to illustrate the proposed procedure. Absolute bias, mean square error, asymptotic confidence intervals, confidence width and coverage probability are obtained. Simulation results indicate that the mean square error is smaller and confidence width is narrower and more precise when number of removals gets smaller. Also, increasing the terminal point of the experiment results in reducing the mean square error and confidence width.

The odd inverse Pareto-Weibull distribution is introduced as a new lifetime distribution based on the inverse Pareto and the T-X family. Some mathematical properties of the new distribution are studied. The method of maximum likelihood is used for estimating the model parameters and the observed Fisher’s information matrix is derived. The importance and flexibility of the proposed model are assessed using a real data.