A SGP data set contains student test records grouped by content areas. It also includes teacher information as well as classroom details where tests were administered, so analyses can easily show student growth within any specific content area. SGP data sets are designed so they are easy to use for analysis.
At one time, most SGP analyses required manual calculations of percentiles by hand for every student and teacher at their respective schools – this took time and could sometimes prove inaccurate. Now however, the SGPdata package provides several functions which automate these calculations – lower level functions (studentGrowthPercentiles and studentGrowthProjections) require data in WIDE format while higher level functions (wrappers for lower level functions) work with either WIDE or LONG data formats.
The SGPdata package provides four example data sets designed to be used with SGP analyses. Each of these serves as a starting point when developing your own analysis using this technique, with its own documentation that details its creation process and analysis applications – it also includes instructions on how to use each function effectively and details about what happens if an error arises.
Gaussian Process regression models tend to work best with smaller to mid-sized datasets due to their high computational costs associated with inverting K. This cost prevents GPR models from being applied on larger data sets as it could cause your machine to slow down when running the regression model.
As it is essential that the data you are using is correct and up-to-date, selecting an accurate site that updates frequently will help ensure accurate data. Any such site will display an indicator next to its name stating when they last updated it; this allows users to avoid problems caused by data that has become out-of-date while simultaneously showing changes from previous updates.