How to automatically inserting default key in prais winsten estimation? - r

Question

I got an error message saying
Error in prais_winsten(agvprc.lm1, data = data) :
argument "index" is missing, with no default
How can I avoid inputting all the features
agvprc.lm1=lm(log(avgprc) ~ mon+tues+wed+thurs+t+wave2+wave3)
summary(agvprc.lm1)
agvprc.lm.pw = prais_winsten(agvprc.lm1, data=data)
summary(agvprc.lm.pw)

Answer 1

Not sure if I've understood the question correctly, but to avoid the "Error in prais_winsten(agvprc.lm1, data = data) : argument "index" is missing, with no default" error you need to provide an 'index' to the function (which is a character variable of "ID" and "time"). Using the inbuilt mtcars dataset as an example, with "cyl" as "time":
library(tidyverse)
#install.packages("prais")
library(prais)
#> Loading required package: sandwich
#> Loading required package: pcse
#>
#> Attaching package: 'pcse'
#> The following object is masked from 'package:sandwich':
#>
#> vcovPC
ggplot(mtcars, aes(x = cyl, y = mpg, group = cyl)) +
geom_boxplot() +
geom_jitter(aes(color = hp), width = 0.2)
agvprc.lm1 <- lm(log(mpg) ~ cyl + hp, data = mtcars)
summary(agvprc.lm1)
#>
#> Call:
#> lm(formula = log(mpg) ~ cyl + hp, data = mtcars)
#>
#> Residuals:
#> Min 1Q Median 3Q Max
#> -0.35699 -0.09882 0.01111 0.11948 0.24118
#>
#> Coefficients:
#> Estimate Std. Error t value Pr(>|t|)
#> (Intercept) 3.7829495 0.1062183 35.615 < 2e-16 ***
#> cyl -0.1072513 0.0279213 -3.841 0.000615 ***
#> hp -0.0011031 0.0007273 -1.517 0.140147
#> ---
#> Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#>
#> Residual standard error: 0.1538 on 29 degrees of freedom
#> Multiple R-squared: 0.7503, Adjusted R-squared: 0.7331
#> F-statistic: 43.57 on 2 and 29 DF, p-value: 1.83e-09
agvprc.lm.pw <- prais_winsten(formula = agvprc.lm1,
data = mtcars,
index = c("hp", "cyl"))
#> Iteration 0: rho = 0
#> Iteration 1: rho = 0.6985
#> Iteration 2: rho = 0.7309
#> Iteration 3: rho = 0.7285
#> Iteration 4: rho = 0.7287
#> Iteration 5: rho = 0.7287
#> Iteration 6: rho = 0.7287
#> Iteration 7: rho = 0.7287
summary(agvprc.lm.pw)
#>
#> Call:
#> prais_winsten(formula = agvprc.lm1, data = mtcars, index = c("hp",
#> "cyl"))
#>
#> Residuals:
#> Min 1Q Median 3Q Max
#> -0.33844 -0.08166 0.03109 0.13612 0.25811
#>
#> AR(1) coefficient rho after 7 iterations: 0.7287
#>
#> Coefficients:
#> Estimate Std. Error t value Pr(>|t|)
#> (Intercept) 3.7643405 0.1116876 33.704 <2e-16 ***
#> cyl -0.1061198 0.0298161 -3.559 0.0013 **
#> hp -0.0011470 0.0007706 -1.489 0.1474
#> ---
#> Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#>
#> Residual standard error: 0.1077 on 29 degrees of freedom
#> Multiple R-squared: 0.973, Adjusted R-squared: 0.9712
#> F-statistic: 523.3 on 2 and 29 DF, p-value: < 2.2e-16
#>
#> Durbin-Watson statistic (original): 0.1278
#> Durbin-Watson statistic (transformed): 0.4019
Created on 2022-02-28 by the reprex package (v2.0.1)
# To present the index without having to write out all of the variables
# perhaps you could use:
agvprc.lm.pw <- prais_winsten(formula = agvprc.lm1,
data = mtcars,
index = names(agvprc.lm1$coefficients)[3:2])
summary(agvprc.lm.pw)
#>
#> Call:
#> prais_winsten(formula = agvprc.lm1, data = mtcars, index = names(agvprc.lm1$coefficients)[3:2])
#>
#> Residuals:
#> Min 1Q Median 3Q Max
#> -0.33844 -0.08166 0.03109 0.13612 0.25811
#>
#> AR(1) coefficient rho after 7 iterations: 0.7287
#>
#> Coefficients:
#> Estimate Std. Error t value Pr(>|t|)
#> (Intercept) 3.7643405 0.1116876 33.704 <2e-16 ***
#> cyl -0.1061198 0.0298161 -3.559 0.0013 **
#> hp -0.0011470 0.0007706 -1.489 0.1474
#> ---
#> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#>
#> Residual standard error: 0.1077 on 29 degrees of freedom
#> Multiple R-squared: 0.973, Adjusted R-squared: 0.9712
#> F-statistic: 523.3 on 2 and 29 DF, p-value: < 2.2e-16
#>
#> Durbin-Watson statistic (original): 0.1278
#> Durbin-Watson statistic (transformed): 0.4019
NB. there are a number of assumptions here that may not apply to your actual data; with more information, such as a minimal reproducible example, you will likely get a better/more informed answer on stackoverflow