dickoa/causalpairs

R functions to load pairs data and build classifiers in the Causal pairs effect competition in Kaggle

Cause Effect Pairs R package

This repo contains some sample code in R to read pairs data (CEdata_text) for the Cause Effect Pairs Challenge, a machine learning challenged hosted by Kaggle and organized by ChaLearn.

Install the package

require(devtools)
install_github("CausalPairs", "dickoa")

How to use it

The main function in this package is the read_pairs_data

require(CausalPairs)
valid <- read_pairs_data("path_to_data/CEdata_valid_pairs.csv")
train <- read_pairs_data("path_to_data/CEdata_train_pairs.csv")

Normally you should get something like

str(train)
Classes ‘data.table’ and 'data.frame':	20569477 obs. of  3 variables:
 $ SampleID: chr  "train1" "train1" "train1" "train1" ...
 $ A       : num  2092 1143 390 1424 1277 ...
 $ B       : num  5651 4449 4012 6124 7310 ...
 - attr(*, "sorted")= chr "SampleID"
 - attr(*, ".internal.selfref")=<externalptr>

More info

For those not really familiar with data.table object, you can check out the vignettes

vignette("datatable-faq", package = "data.table")
vignette("datatable-intro", package = "data.table")

More ressources at data.table project