Introduction
-spmodel is an package used to fit, summarize, and
-predict for a variety of spatial statistical models. The vignette
-provides an introduction spatial generalized linear models for
+
spmodel is an R package used to fit,
+summarize, and predict for a variety of spatial statistical models. The
+vignette provides an introduction spatial generalized linear models for
non-Gaussian response distributions in spmodel. Before
proceeding, we load spmodel by running
@@ -309,9 +309,10 @@The Spatial Generalized Linear Mod (2024) provide the methodology’s full details.
+binomial and beta response distributions (specified by +
spmodelaccommodates the binomial, beta, Poisson, negative binomial, gamma, and inverse Gaussian distributions. The -binomial and beta response distributions (specified by ) use the logit -link, while the Poisson, negative binomial, gamma, and inverse Gaussian -distributions use the log link – see the following table.family) use the logit link, while the Poisson, negative +binomial, gamma, and inverse Gaussian distributions use the log link – +see the following table.
| \(\texttt{formula}\) | +formula |
Fixed effects formula | Both | ||
| \(\texttt{family}\) | +family |
Response distribution | Both | ||
| \(\texttt{data}\) | -Data Frame or object | +data |
+Data Frame or sf object |
Both | |
| \(\texttt{spcov_type}\) | +spcov_type |
Spatial covariance type | Both | ||
| \(\texttt{spcov_initial}\) | +spcov_initial |
Initial or known spatial covariance parameters | Both | ||
| \(\texttt{dispersion_initial}\) | +dispersion_initial |
Initial or known dispersion parameter | Both | ||
| \(\texttt{estmethod}\) | +estmethod |
Estimation method | Both | ||
| \(\texttt{random}\) | +random |
Random effects formula | Both | ||
| \(\texttt{randcov_initial}\) | +randcov_initial |
Initial or known random effect variances | Both | ||
| \(\texttt{partition_factor}\) | +partition_factor |
A partition factor formula | Both | ||
| \(\texttt{...}\) | +... |
Additional arguments | Both | ||
| \(\texttt{anisotropy}\) | +anisotropy |
Whether to model anisotropy | -\(\texttt{spglm()}\) | +spglm() |
|
| \(\texttt{xcoord}\) | +xcoord |
An x-coordinate name | -\(\texttt{spglm()}\) | +spglm()} |
|
| \(\texttt{ycoord}\) | +ycoord |
A y-coordinate name | -\(\texttt{spglm()}\) | +spglm() |
|
| \(\texttt{local}\) | +local |
Options for large data sets | -\(\texttt{spglm()}\) | +spglm() |
|
| \(\texttt{W}\) | +W |
A neighborhood weight matrix | -\(\texttt{spgautor()}\) | +spgautor() |
|
| \(\texttt{row_st}\) | -Whether to row-standardize \(\texttt{W}\) + | row_st |
+Whether to row-standardize W
|
-\(\texttt{spgautor()}\) | +spgautor() |
| \(\texttt{M}\) | +M |
A symmetry matrix | -\(\texttt{spgautor()}\) | +spgautor() |
|
| \(\texttt{range_positive}\) | +range_positive |
Whether the range parameter is restricted to be positive | |||
| \(\texttt{cutoff}\) | +cutoff |
Distance cutoff if using distance-based neighbor definition |
| Argument | Purpose | ||
|---|---|---|---|
| \(\texttt{object}\) | +object |
Fitted model object | |
| \(\texttt{newdata}\) | -New data frame or object | +newdata |
+New data frame or sf object |
| \(\texttt{type}\) | +type |
Whether predictions are on the link or response scale | |
| \(\texttt{se.fit}\) | +se.fit |
Whether standard errors should be returned | |
| \(\texttt{interval}\) | +interval |
Interval type (prediction or confidence) | |
| \(\texttt{level}\) | +level |
Prediction or confidence level | |
| \(\texttt{local}\) | +local |
Options for large data sets | |
| \(\texttt{var_correct}\) | +var_correct |
Whether prediction variances should be adjusted | |
| \(\texttt{...}\) | +... |
Additional arguments |
