Optimal and Coherent Data Visualization in R for the Empirical Study of CPI‐U Standard Errors

Harold Gomes

Abstract

This study looks at the density distribution of standard errors (SE) of item stratum-index area level percent change using Consumer Price Index for All Urban Consumers (CPI-U) historical data in order to understand anomalous behavior. Many attributes can be determined about the SE, such as: the shape of the SE density distribution; whether the overall central tendency and overall variability of the SE distribution are smaller, larger, or similar from one year to the next, or from one month to the same month of the following year; whether the distributions tend to shift over time or stay stationary. The SE of basic level price changes are used to produce the underlying distributions for further examination. SE for May 2006 in particular were investigated as it was the month when the SE of 12-month CPI-U percent change reached its largest value of 0.19. Non-parametric methods and categorical analysis techniques were employed to assess the May 2004-May 2008 datasets. A compelling approach to data visualization is produced by combining multiple pieces of information in a single graph in order to demonstrate an optimal and coherent visual of comparisons.