Subfolders contain the results of the final spatial subsampling assessments of considered indicator variables. Tests are described in the report.
Within each folder:
-Each filename consists of a header (e.g., "BR_BS_20170323b") which refers to the province ("BR"), date ("20170323"), and run ("b") of the spatial subsample run ("BS": "BootStrap").
-"_FitStats" is a table of summary statistics
-_"TTestStats" is a table of T-test summary statistics
-"Subsamples" subfolder contains the 100 text file data tables that were analyzed for the spatial subsampling assessment
-Remaining files are summary coefficient evaluation plots. Names refer to various indicator parameters.
-------------------------------
For "FitPlots" and in the statistics tables, indicator parameter names are as follows:
"AspectAspectBin__1": Slope aspect category -1 (flat)
"AspectAspectBin_2": Slope aspect category 2 (south)
NOTE: Slope aspect category 1 (north) was used as a reference category for the slope aspects
"CurvNegBin": The binned negative curvature continuous parameter
"CurvPosBin": The binned positive curvature continuous parameter
"PRISMBin": The binned precipitation data from the PRISM Climate Group, Oregon State University
"SlopeBin": The binned slope angle continuous parameter
"FaultsBuffer": The proximity-to-fault parameter (value of 1 if within 1 km of a fault on the statewide geologic map, 0 otherwise)
"RiversBuffer": The proximity-to-major river floodplain parameter (value of 1 if within 500 m of a major river floodplain, 0 otherwise)
"Geology": Geology categorical variable. The categories vary with province.
NOTE: GeologyS1 (or GeologySu, in the Mogollon-Datil area) was used as a reference category for the geology categories
A parameter missing from either the statistics tables or from the FitPlots list means the parameter was not sample adequately enough to provide statistics.
-------------------------------
Table abbreviations:
FitStats table:
"Num.In" - Number of subsample results for which the R glm() function returned coefficient estimates
"Num.Outliers" - Number of subsample results with coefficent estimate and coefficient estimate standard deviations outside the 1.5*interquartile range outlier criteria
"Num.LargeStDev" - Number of subsample results with coefficient standard deviation greater than the range of coefficient estimates
"Num.Fails.PVal" - Number of subsample results with Wald test p-value <0.01
"SW_Coeffs.Stat" - Shapiro-Wilk normality test statistic
"SW_Coeffs.PVal" - Shaprio-Wilk normality test p-value
"NFit_Coeffs.Mean" - Mean of a normal curve fit to the distribution of coefficients
"NFit_Coeffs.SD" - Standard deviation of a normal curve fit to the distribution of coefficients
"NFit_Coeffs.KS.Stat" - Kolmogorov-Smirnov one-sample test statistic assessing goodness-of-fit for the normal curve fit to the distribution of coefficients
"NFit_Coeffs.KS.PVal" - Kolmogorov-Smirnov one-sample test statistic p-value assessing goodness-of-fit for the normal curve fit to the distribution of coefficients
"Range.StDeviations" - Range in standard deviations of subsample coefficient estimates
TTestStats table:
T-test performed to test if mean of distribution of coefficients is <>0
"T.PVal" - T-test p-value
"T.DF" - T-test degrees-of-freedom
"T.Mean" - mean of the distribution of coefficients
------------------------------
Fit Plots abbreviations:
"S-W p-val": Shapiro-Wilk normality test p-value
"K-S p-val": Kolmogorov-Smirnov goodness-of-fit test p-value for the fit of the Gaussian to the distribution of coefficients
"Positive:Negative": Ratio of positive coefficient estimates to negative coefficient estimates
Individual subsample coefficients are plotted as small purple circles along the x-axis, while the uncertainty in each is represented by a small green Gaussian centered on the estimate with standard deviation equal to the standard deviation in the coefficient estimate from the R glm() function.