I have a data frame (out.predict.garch) in, what I want to get is to filter the data following the condition that the variables out.predict.garch $ name is equal to one of the levels (ibex) (that if it is done) and that the variable out.predict.garch $ error is the smallest value, thus staying with a single observation in the new data frame.
I have found several formulas for filtering data frame, subset, select, filtrer, but none of them find the lower value of a numeric variable.
Without a reproducible example it is difficult to respond. There are several ways to do it, one option could be using the function which.min() :
out.predict.garch.ibex <- out.predict.garch[out.predict.garch$nombre=='ibex', ]
out.predict.garch.ibex[which.min(out.predict.garch.ibex$error), ]
I hope you solve your problem.
Without a doubt the answer of P. Paccioretti is the appropriate one, but in any case I offer you an approximation to the problem using the package included in tidyverse . Using the pipe %>% plus the data manipulation verbs, once you get used to it, it can be very practical, performant and tremendously clear / readable.
Your problem could be solved using the "verb" filter :
library(tidyverse)
out.predict.garch %>% # a partir de out.predict.garch
filter(nombre == "DIA", # filtramos por nombre == "DIA"
error == min(error)) # y por aquellos dónde tenemos el mínimo error