Statistics

The Explorations in Statistics Research workshop is a one-week NSF-funded summer program that introduces undergraduate students to current research problems in applied statistics. The goal of the workshop is to expose students to exciting, modern applied statistical research and practice, with the ultimate aim of interesting them in seeking more training in statistics at the undergraduate and graduate levels. The program is explicitly designed to engage students in the connections between authentic domain problems and the statistical ideas and approaches needed to address these problems, which is an important aspect of statistical thinking that is difficult to teach and sometimes lacking in our methodological courses and programs. Over the past nine years, we ran the workshop six times and a similar program in the sciences two times. We describe the program, summarize feedback from participants, and identify the key features to its success. We abstract these features and provide a set of recommendations for how faculty can incorporate important elements into their regular courses.

This work investigates whether and how COVID-19 containment policies had an immediate impact on crime trends in Los Angeles. The analysis is conducted using Bayesian structural time-series and focuses on nine crime categories and on the overall crime count, daily monitored from January 1st 2017 to March 28th 2020. We concentrate on two post-intervention time windows - from March 4th to March 16th and from March 4th to March 28th 2020 - to dynamically assess the short-term effects of mild and strict policies. In Los Angeles, overall crime has significantly decreased, as well as robbery, shoplifting, theft, and battery. No significant effect has been detected for vehicle theft, burglary, assault with a deadly weapon, intimate partner assault, and homicide. Results suggest that, in the first weeks after the interventions are put in place, social distancing impacts more directly on instrumental and less serious crimes. Policy implications are also discussed.

In this note we discuss the mathematical tools to define trend indicators which are used to describe market trends. We explain the relation between averages and moving averages on the one hand and the so called exponential moving average (EMA) on the other hand. We present a lot of examples and give the definition of the most frequently used trend indicator, the MACD, and discuss its properties.

In this paper, we introduce a new class of distributions by compounding the exponentiated extended Weibull family and power series family. This distribution contains several lifetime models such as the complementary extended Weibull-power series, generalized exponential-power series, generalized linear failure rate-power series, exponentiated Weibull-power series, generalized modified Weibull-power series, generalized Gompertz-power series and exponentiated extended Weibull distributions as special cases. We obtain several properties of this new class of distributions such as Shannon entropy, mean residual life, hazard rate function, quantiles and moments. The maximum likelihood estimation procedure via a EM-algorithm is presented.

We develop a Bayesian framework for estimation and prediction of dynamic models for observations from the two-parameter exponential family. Different link functions are introduced to model both the mean and the precision in the exponential family allowing the introduction of covariates and time series components. We explore conjugacy and analytical approximations under the class of partial specified models to keep the computation fast. The algorithm of West, Harrison and Migon (1985) is extended to cope with the two-parameter exponential family models. The methodological novelties are illustrated with two applications to real data. The first, considers unemployment rates in Brazil and the second some macroeconomic variables for the United Kingdom.

Extreme shock models have been introduced in Gut and Hüsler (1999) to study systems that at random times are subject to shock of random magnitude. These systems break down when some shock overcomes a given resistance level. In this paper we propose an alternative approach to extreme shock models using reinforced urn processes. As a consequence of this we are able to look at the same problem under a Bayesian nonparametric perspective, providing the predictive distribution of systems' defaults.

Bertand's paradox is a fundamental problem in probability that casts doubt on the applicability of the indifference principle by showing that it may yield contradictory results, depending on the meaning assigned to "randomness". Jaynes claimed that symmetry requirements (the principle of transformation groups) solve the paradox by selecting a unique solution to the problem. I show that this is not the case and that every variant obtained from the principle of indifference can also be obtained from Jaynes' principle of transformation groups. This is because the same symmetries can be mathematically implemented in different ways, depending on the procedure of random selection that one uses. I describe a simple experiment that supports a result from symmetry arguments, but the solution is different from Jaynes'. Jaynes' method is thus best seen as a tool to obtain probability distributions when the principle of indifference is inconvenient, but it cannot resolve ambiguities inherent in the use of that principle and still depends on explicitly defining the selection procedure.

Predictive maintenance and condition-based monitoring systems have seen significant prominence in recent years to minimize the impact of machine downtime on production and its costs. Predictive maintenance involves using concepts of data mining, statistics, and machine learning to build models that are capable of performing early fault detection, diagnosing the faults and predicting the time to failure. Fault diagnosis has been one of the core areas where the actual failure mode of the machine is identified. In fluctuating environments such as manufacturing, clustering techniques have proved to be more reliable compared to supervised learning methods. One of the fundamental challenges of clustering is developing a test hypothesis and choosing an appropriate statistical test for hypothesis testing. Most statistical analyses use some underlying assumptions of the data which most real-world data is incapable of satisfying those assumptions. This paper is dedicated to overcoming the following challenge by developing a test hypothesis for fault diagnosis application using clustering technique and performing PERMANOVA test for hypothesis testing.

This study is to investigate the feasibility of least square method in fitting non-Gaussian noise data. We add different levels of the two typical non-Gaussian noises, Lévy and stretched Gaussian noises, to exact value of the selected functions including linear equations, polynomial and exponential equations, and the maximum absolute and the mean square errors are calculated for the different cases. Lévy and stretched Gaussian distributions have many applications in fractional and fractal calculus. It is observed that the non-Gaussian noises are less accurately fitted than the Gaussian noise, but the stretched Gaussian cases appear to perform better than the Lévy noise cases. It is stressed that the least-squares method is inapplicable to the non-Gaussian noise cases when the noise level is larger than 5%.

The occurrence of digits 1 through 9 as the leftmost nonzero digit of numbers from real-world sources is distributed unevenly according to an empirical law, known as Benford's law or the first digit law. It remains obscure why a variety of data sets generated from quite different dynamics obey this particular law. We perform a study of Benford's law from the application of the Laplace transform, and find that the logarithmic Laplace spectrum of the digital indicator function can be approximately taken as a constant. This particular constant, being exactly the Benford term, explains the prevalence of Benford's law. The slight variation from the Benford term leads to deviations from Benford's law for distributions which oscillate violently in the inverse Laplace space. We prove that the whole family of completely monotonic distributions can satisfy Benford's law within a small bound. Our study suggests that Benford's law originates from the way that we write numbers, thus should be taken as a basic mathematical knowledge.