# Research

## Frontiers in Statistics

### Friday, April 21, 2006

Title: Distributions, moments, inference problems, Jordan Stoyanov
Speaker: Jordan Stoyanov, Newcastle University, UK
Time: 3:00pm‐4:00pm
Place: PHY 118

#### Abstract

The main discussion will be on distributions and their properties expressed in terms of the moments. It will be clear how important is the distributions we deal with to be uniquely determined by their moments.

### Friday, March 31, 2006

Title: Statistical Analysis of Lightning Data
Speaker: Rebecca Wooten
Time: 3:00pm‐4:00pm
Place: PHY 118

### Friday, March 24, 2006

Title: Analyzing lightning data using records
Speaker: Alfred Mbah
Time: 3:00pm‐4:00pm
Place: PHY 118

#### Abstract

We show by simulation that results obtained using record breaking data are as good as the results obtained using the entire sample of size n.

We use records to analyze lightning data.

### Friday, February 24, 2006

Title: Logistic Regression Approach to Software Reliability Assessment: Early Estimation of Parameters
Speaker: Louis Camara
Time: 3:00pm‐4:00pm
Place: PHY 118

#### Abstract

While modeling software reliability data, is a topic of major importance that has useful industrial applications, an early estimation of the number of fault in the software would be very beneficial to the software developers. Assuming the logistic model, an effective procedure for estimating the number of faults in a software early in the testing and debugging phase will be presented. Using real software failure data, we will illustrate the effectiveness of our results.

### Friday, February 17, 2006

Title: Correlation of Storm Characteristics with Constituent Concentration in Urban Storm Water Discharges
Speaker: L. Donald Duke, Ph.D., P.E.
Department of Environmental Science and Policy, USF
Time: 3:00pm‐4:00pm
Place: PHY 118

#### Abstract

This research quantifies the correlation between characteristics of storm events and the event mean concentration (EMC) of selected chemically-conservative constituents in runoff originating from those events, using seven urban watersheds in semi-arid coastal California cities. It features a method that employs normality testing to identify extreme storm events, where EMCs have a different mathematical relationship with storm characteristics than is the case with other events. Removing extreme events provides a somewhat better correlation.

### Friday, February 10, 2006

Title: On Simple Branching Processes that Grow Faster than Complete N-ary Trees
Speaker: George Yanev
Time: 3:00pm‐4:00pm
Place: PHY 118

#### Abstract

Branching processes are individual-based stochastic models for the growth of populations. They have important applications in biology and epidemiology, among others. What is the probability that a branching process grows faster than a complete binary tree? What is the critical reproduction rate that makes this probability positive? What is the distribution of the number of complete N-ary subtrees of a branching tree? We will discuss the answers to these questions as well as some open problems.

### Friday, January 27, 2006

Title: On Characterizing Distributions with Conditional Expectations of Functions of Generalized Order Statistics
Speaker: M. I. Beg, Visiting Professor
(joint work with M. Ahsanullah)
Time: 3:00pm‐4:00pm
Place: PHY 118

#### Abstract

Let $$X(1,n,m,k)$$, $$X(2,n,m,k)$$, $$X(n,n,m,k)$$ be $$n$$ generalized order statistics from an absolutely continuous distribution. We give characterizations of distributions by means of $$E\{\psi(X(s,n,m,k))\mid X(r,n,m,k)=x\}=g_1(x)$$ and $$E\{\psi(X(r,n,m,k))\mid X(s,n,m,k)=x\}=g_2(x)$$, $$s>r$$ under some mild conditions on $$\psi(.)$$, $$g_i(x)$$, $$i=1,2$$. It is shown that most of the known characterization results based on conditional expectations are special cases of the results of this paper.

### Friday, January 20, 2006

Title: Analysis of Data from Response Guided Multiple-Baseline Designs
Speaker: John Ferron, USF College of Education
Time: 3:00pm‐4:00pm
Place: PHY 118

#### Abstract

Multiple-baseline designs are frequently used in educational contexts to make treatment effect inferences. Multiple-baseline designs typically lead to the collection of interrupted time series data on three to five participants. The inferences are often drawn from short series lengths (less than 20 observations) that arise from response guided experimentation (using the observed data to guide decisions about how much data to collect before and after intervention). An analysis of response guided multiple-baseline data will be presented.