strm: strm: Spatio-temporal regression model

In strm: Spatio-Temporal Regression Modeling

Description

The strm function provides maximum likelihood estimation of a spatio-temporal simultaneous autoregressive lag error model. This package is built on the errorsarlm() function from the spatialreg package.

Usage

strm(
  formula,
  id,
  data,
  listw,
  time = 2,
  wide = FALSE,
  filter_options = NULL,
  returndf = FALSE,
  ...
)

Arguments

formula	Model formula specified by user (without lags). Any transformed variables, such as logged-variables, should be specified in the model formula.
id	Group identifier (example: state).
data	Name of dataframe.
listw	Spatial weights list object.
time	Number of time periods in the dataset. Lags will be taken for each time period. Default is 2 time periods. For a spatial-only regression model, set time=1.
wide	Boolean indicator. Takes TRUE if data is in wide format and FALSE if data is in long format. If data is in wide format, it is assumed that the user is including the temporal lags for the explanatory variables and response variable manually. Default is FALSE.
filter_options	Additional arguments to be passed to dplyr::filter(). Default is NULL.
returndf	Logical. If modified dataframe should be returned. Default is FALSE.
...	Additional arguments to be passed to spatialreg::errorsarlm().

Details

Spatio-temporal regression model

Any transformed variables should be included in the formula statement. For example, to request gdp natural log-transformed, you would build the model formula as log(gdp).

Examples

library(spdep)
library(dplyr)
data("Produc", package="Ecdat")
data("usaww")
usalw <- mat2listw(usaww)
formula <- as.formula( log(gsp)  ~ log(pcap) + log(pc) + log(emp) + unemp)
out1 <- strm(formula, id="state", data=Produc, 
listw = usalw, time=2,wide=FALSE,
filter_options="year==1970 | year==1971")
out2 <- strm(formula, id="state", data=Produc, 
listw= usalw, time=2, wide = FALSE, 
filter_options="year==1970 | year==1971",
 method="Chebyshev")

strm documentation built on Jan. 19, 2022, 9:06 a.m.