Justin R. Chimka

An Online RFID Laboratory Learning Environment

Due to the increasing demand for RFID expertise and the existence of a knowledge gap between industry and academia in this domain, work has been stimulated to help spread understanding in this field and bridge the gap between theoretical examinations and industrial practices. Among the encouraged work, there is the Integrated Auto-ID Technology for Multidisciplinary Undergraduate Studies (I-ATMUS) project that involved developing an online learning environment for RFID technology with a remotely controllable laboratory system. Technological resources can now be accessed by learners through the Web technology to apply appropriate configurations to the system, conduct experiments using RFID technology, and perform statistical analysis on the acquired data. The developed educational tool was used by two sets of students, who showed improvements in their confidence, knowledge, and skills.

Proportional Hazards Models of Graduation.

Survival analysis is a statistical tool used to describe the duration between events. Many processes in medical research, engineering, and economics can be described using survival analysis techniques. This research involves studying engineering college student graduation using Cox proportional hazards models. Among male students with American College Test (ACT) scores only, those with better Science ACT scores were more likely to graduate. Among students submitting only Scholastic Assessment Test (SAT) scores upon application, controlling for other descriptors such as in-state residence, hometown population, and student major, students with better SAT Math score and female students were more likely to graduate. In summary, results reported here are the significance of standardized math test scores, gender, and Science ACT scores in explaining variation in student graduation under different conditions.

Fuzzy reliability in conceptual design

Conceptual designs intrinsic uncertainty and influence on product life cycle cost make reliability engineering tools during early design phases important. However, typically poor knowledge bases make the prediction of component failure and system performance challenging. Fuzzy techniques are used here during initial conceptual design stages to quantify imprecision and uncertainty in reliability and risk analysis. More specifically, a fuzzy approach is compared to the probabilistic approach presented in Reliability prediction models to support conceptual design by Ormon, et al. (2002). Triangular fuzzy numbers instead of triangular probability distributions represent unknown failure rates here. Two examples of fuzzy reliability analysis in conceptual design are presented where system reliability is evaluated at the subsystem level: The first is to familiarize the reader with fuzzy reliability subsystem level analysis. The second demonstrates fuzzy reliability prediction models for conceptual design tradeoffs. Examples include subsystems operating with active and standby redundancy. In the first example, defuzzified system reliability is a more conservative prediction than that found probabilistically. Recommendations in Example 2 do not differ from those based on the probabilistic models of Ormon, et al.

Type II errors of demerit control charts

ABSTRACT Complex products may present more than one type of defect. A demerit control chart is a useful tool for monitoring different types of defects in a single chart while taking into account different levels of severity. Traditionally, control limits have been established based on standard deviations from the centerline assuming normality. These limits were improved upon by approaches to finding the exact distribution of the demerit statistic and establishing probability-based limits. With respect to Type I error, probability-based limits have been shown to outperform traditional limits. Now, again with exact distributions, we consider Type II errors as well when establishing control limits for different shifts, means and weights.

Statistical quality control for DNA microarray data: A model of type I error

ABSTRACT Genetics and biotechnology have created interesting opportunities for statistical quality control research into DNA microarray data. We describe here a study of gene expression data subjected to control charts for individual measurements and Hotelling T 2 control charts, where observed errors are modeled as a function of chart and varied decision rules applied to reject the hypothesis of control. Results suggest that while a good metamodel of Type I error exists, microarray image processing and normalization to remove systematic variation would seem inefficient. And additional work is necessary to develop more useful quality technology for DNA microarrays.

Read rate analysis of radio frequency identification systems for business applications

This paper presents a study on operational effectiveness of Radio Frequency Identification (RFID) systems in a controlled environment. Logistic regression models are developed to capture the dynamic relationship among implementation factors affecting the performance of an RFID system. Distance and angle between tag and antenna, orientation and position of tag on the container are found to be important factors among other implementation factors. A test bed is used to represent the flow of tagged products on a rolling conveyor passing by an antenna. It is shown that controlled and designed experimentation about RFID yields valuable data that, subject to appropriate statistical models, may result in a better practical understanding of the technology. In addition to the performance assessment and evaluation for RFID systems with respect to various factors, this study also aims to establish a reliable literature base for the development of this technology in the future.

Improved (and Singular) Disinfectant Protocol for Indirectly Assessing Organic Precursor Concentrations of Trihalomethanes and Dihaloacetonitriles

Measurements of disinfection byproduct (DBP) organic precursor concentrations (OPCs) are crucial to assess and improve DBP control processes. Typically, formation potential tests - specified in Standard Methods (SM) 5710-B/D - are used to measure OPCs. Here, we highlight several limitations of this protocol for dihaloacetonitriles and trihalomethanes and validate a novel Alternative Method (AM). The effects of pH, disinfectant type (free chlorine and monochloramine), and chlor(am)ine residual (CR) were examined on DBP formation in a suite of waters. Using the SM, DHAN decreased 43-47% as the CR increased from 3 to 5 mg L(-1) as Cl2, compromising OPC assessments. In contrast, a high monochloramine dose (250 mg L(-1) as Cl2) at pH 7.0 (the AM) accurately reflected OPCs. The two methods were compared for assessing DBP precursor removal through three granular activated carbon (GAC) columns in series. Breakthrough profiles assessed using the AM only showed DBP precursor sorption occurred in each column that decreased over time (p = 0.0001). Similarly, the AM facilitated ranking of three types of GAC compared in parallel columns, whereas the SM produced ambiguous results. Fluorescence intensity of a humic-like fluorophore (i.e., I345/425) correlated strongly to precursor removal in the GAC columns. The practical implications of the results are discussed.

Optimizing the Sunday Singles Lineup for a Ryder Cup Captain

The Ryder Cup competition between Team USA and Team Europe is the premier team event in professional golf, largely because of the drama associated with head-to-head matches on the final day (Sunday). The captain of each team must determine the exact sequence of his 12 golfers in the 12 Sunday singles matches, and neither captain is precisely aware of the other captains intended lineup. Because the losing captain has often been criticized for choosing what the critics deem to be a poor lineup, we develop a quantitative methodology that a captain can use to determine his Sunday singles lineup to maximize the probability that his team is in possession of the Ryder Cup upon completion of the competition. As part of our effort, we develop a statistical model for estimating the three possible outcomes (win, lose, tie) of each singles match. Using the 1989–2008 Ryder Cup competitions as examples, we find that using our statistical model within our methodology indicates that the Sunday singles competition is not greatly impacted by the lineup selected by the captain. However, we define another example that indicates our methodology can be of great benefit if a players performance is affected by who his singles match opponent is.

Gamma Regressive Individuals Control Charts for Influenza Activity

ABSTRACT Statistical process control has been considered along with generalized linear models of influenza activity. Different combinations of distribution and link assumptions have been formally compared in order to choose one that would seem theoretically appropriate. Specificity of gamma regression–based individuals control charts for influenza activity has been estimated and compared to begin answering the important question: What are reasonable expectations about errors associated with outbreak detection? Results include performance among gamma regression–based charts for laboratory data that is relatively near what one should expect from traditional process control. It is shown how more theoretically appropriate statistically quality control models of flu activity become increasingly specific and so useful in identifying non-seasonal outbreaks. And it is shared how this research is being applied in practice.

Modeling reliability with a two-sided power distribution

ABSTRACT Modeling the reliability of a system typically involves the piecewise combination of mathematically disparate functions covering distinct periods (e.g., decreasing, constant, and increasing time-dependent failure rates defining the traditional bathtub curve) in the life of an engineered system. This work explores the two-sided power distribution, a nonlinear extension of the triangular distribution, for describing failure time and rendering a statistically valid bathtub curve with a single probability density function. The parameters of the two-sided power distribution also provide some flexibility to describe decreasing and increasing failure rates with a single distribution that are not available with similar distributions.