mrdomani / autofeat

Scalable Automatic Feature Engineering

Scalable Automatic Feature Engineering with autofeat package

Generate automatically new features based on older ones for further modelling, using SAFE algoritm proposed in a paper by Shi, Zhang, Li, Yang and Zhou. This is a direct implementation of the pseudo-algoritm proposed in the paper, with its conventions, denotements and flaws.

Install

# install.packages("remotes")
remotes::install_github("MrDomani/autofeat")

autofeat's Issues

Does this package have a data leakage problem?

Hi! I'm interested in this package, and want to use this package for my project. However, it seems that this package uses both training set and testing set during the feature construction process. Won't this result in an over-fitting problem?

Error: ! Arguments in `...` must be passed by position, not name.

set.seed(1)
X <- matrix(rnorm(2000),ncol = 5)
colnames(X) <- paste0("V",1:ncol(X))

Y <- sample(0:1,size = nrow(X),replace = T) |> as.factor()

tr <- 1:200
ts <- 201:nrow(X)

library(autofeat)
s <- SAFE(X_train = X[tr,],
          y_train = Y[tr],
          X_valid = X[ts,],
          y_valid = Y[ts])

Error:
! Arguments in `...` must be passed by position, not name.
x Problematic arguments:
* V2 = <fct>
* V3 = <fct>
Run `rlang::last_trace()` to see where the error occurred.