6 validate()
The glance() function from broom had a vague label for the F statistic (simply “statistic”) and lacked any kind of pseudo R-squared for logistic regressions.
Furthermore, while the same function is friendly for data frames, its wide form is cumbersome for quickly ascertaining model validity. Thus, validate() produces similar output as a column vector, adding McFadden’s pseudo R-squared and the apparent error rate–defined as the ratio of the number of incorrect predictions to correct ones (i.e. number incorrect / number correct)–for logistic regressions. Those who wish to have the values in the format of broom can always transpose the vector. Alternatively, converting the output to a dataframe is simple by setting dataframe = TRUE in the function.
Output definitions are in the help file associated with this function.
6.1 Case 1: OLS
## model.lm
## n 32.000000
## rsq 0.753842
## adj.rsq 0.736866
## F.stat 44.405361
## df.num 3.000000
## df.den 29.000000
## p.value 0.000000
## residual.median -0.293202
## residual.mean 0.000000
## residual.sd 2.990226
## residual.se 0.528602
## rmse 2.943133
## mad 1.943778
## mae 2.353567
## medianpe -0.016107
## mpe -0.015267
## sdpe 0.161915
## sepe 0.028623
## AIC 167.898446
## BIC 173.761389
## loglik -79.949223## statistic model.lm
## 1 n 32.000000
## 2 rsq 0.753842
## 3 adj.rsq 0.736866
## 4 F.stat 44.405361
## 5 df.num 3.000000
## 6 df.den 29.000000
## 7 p.value 0.000000
## 8 residual.median -0.293202
## 9 residual.mean 0.000000
## 10 residual.sd 2.990226
## 11 residual.se 0.528602
## 12 rmse 2.943133
## 13 mad 1.943778
## 14 mae 2.353567
## 15 medianpe -0.016107
## 16 mpe -0.015267
## 17 sdpe 0.161915
## 18 sepe 0.028623
## 19 AIC 167.898446
## 20 BIC 173.761389
## 21 loglik -79.9492236.2 Case 2: GLM (logit)
model.glm <- glm(formula = am ~ mpg + wt, mtcars,
family = binomial(link = 'logit'))
validate(model.glm) ## model.glm
## n 32.000000
## pseudo.rsq.mcfad 0.602490
## aer 0.062500
## null.deviance 43.229733
## residual.deviance 17.184255
## df.null 31.000000
## df.residual 29.000000
## residual.median -0.046842
## residual.mean -0.044152
## residual.sd 0.743181
## residual.se 0.131377
## rmse 0.732808
## mad 0.384793
## mae 0.508942
## medianpe 1.024399
## mpe 0.881613
## sdpe 0.482775
## sepe 0.085343
## AIC 23.184255
## BIC 27.581463
## loglik -8.5921286.3 Case 3: NLS
## model.nls
## n 116.000000
## iterations 4.000000
## convergence_tolerance 0.000001
## sigma 23.624178
## df.sigma 114.000000
## residual.median -0.684547
## residual.mean -0.000002
## residual.sd 23.521240
## residual.se 2.183892
## rmse 23.419636
## mad 15.047691
## mae 17.120045
## medianpe -0.011579
## mpe -0.287614
## sdpe 1.161944
## sepe 0.107884
## AIC 1066.823097
## BIC 1075.083868
## loglik -530.411549