Dynamic Graphics and Reporting for Statistics

Abstract

<p>Statistics as a scientific discipline has a dynamic nature, which can be</p> <p>observed in many statistical algorithms and theories as well as in data</p> <p>analysis. For example, asymptotic theories in statistics are inherently</p> <p>dynamic: they describe how a statistic or an estimator behaves as the sample</p> <p>size increases. Data analysis is almost never a static process. Instead, it</p> <p>is an iterative process involving cleaning, describing, modeling, and</p> <p>re-cleaning the data. Reports may end up being re-written due to changes in</p> <p>the data and analysis.</p> <p>This thesis consists of three parts, addressing the dynamic aspects of</p> <p>statistics and data analysis. In the first part, we show how to explain the</p> <p>ideas behind some statistical methods using animations, followed by an</p> <p>introduction to the design and functionality of the animation package. In</p> <p>the second part, we discuss the design of an interactive statistical</p> <p>graphics system, with an emphasis on the reactive programming paradigm and</p> <p>its connection with the data infrastructure in R, as utilized in the cranvas</p> <p>package. In the third part, we provide a solution to statistical reporting,</p> <p>which is implemented in the knitr package, making use of literate</p> <p>programming. It frees us from the traditional approach of cut-and-paste, and</p> <p>provides a seamless integration of computing and reporting that enhances</p> <p>reproducible research. Demos and examples were given along with the</p> <p>discussion.</p>