Site icon R-bloggers

Linear model, xgboost and randomForest cross-validation using crossval::crossval_ml

[This article was first published on T. Moudiki's Webpage - R, and kindly contributed to R-bloggers]. (You can report issue about the content on this page here)
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

As seen last week in a post on grid search cross-validation, crossval contains generic functions for statistical/machine learning cross-validation in R. A 4-fold cross-validation procedure is presented below:

In this post, I present some examples of use of crossval on a linear model, and on the popular xgboost and randomForest models. The error measure used is Root Mean Squared Error (RMSE), and is currently the only choice implemented.

Installation

From Github, in R console:

devtools::install_github("thierrymoudiki/crossval")

Demo

We use a simulated dataset for this demo, containing 100 examples, and 5 explanatory variables:

# dataset creation
 set.seed(123)
 n <- 100 ; p <- 5
 X <- matrix(rnorm(n * p), n, p)
 y <- rnorm(n)

Linear model

Linear model example:

crossval::crossval_ml(x = X, y = y, k = 5, repeats = 3)

## $folds
##         repeat_1  repeat_2  repeat_3
## fold_1 0.8987732 0.9270326 0.7903096
## fold_2 0.8787553 0.8704522 1.2394063
## fold_3 1.0810407 0.7907543 1.3381991
## fold_4 1.0594537 1.1981031 0.7368007
## fold_5 0.7593157 0.8913229 0.7734180
## 
## $mean
## [1] 0.9488758
## 
## $sd
## [1] 0.1902999
## 
## $median
## [1] 0.8913229

Linear model example, with validation set:

crossval::crossval_ml(x = X, y = y, k = 5, repeats = 3, p = 0.8)

## $folds
##                    repeat_1  repeat_2  repeat_3
## fold_training_1   1.1256933 0.9144503 0.9746044
## fold_validation_1 0.9734644 0.9805410 0.9761265
## fold_training_2   1.0124938 0.9652489 0.7257494
## fold_validation_2 0.9800293 0.9577811 0.9631389
## fold_training_3   0.7695705 1.0091999 0.9740067
## fold_validation_3 0.9753250 1.0373943 0.9863062
## fold_training_4   1.0482233 0.9194648 0.9680724
## fold_validation_4 0.9984861 0.9596531 0.9742874
## fold_training_5   0.9210179 1.0455006 0.9886350
## fold_validation_5 1.0126038 0.9658146 0.9658412
## 
## $mean_training
## [1] 0.9574621
## 
## $mean_validation
## [1] 0.9804529
## 
## $sd_training
## [1] 0.1018837
## 
## $sd_validation
## [1] 0.02145046
## 
## $median_training
## [1] 0.9740067
## 
## $median_validation
## [1] 0.975325

Random Forest

randomForest example:

require(randomForest)

# fit randomForest with mtry = 4

crossval::crossval_ml(x = X, y = y, k = 5, repeats = 3,
fit_func = randomForest::randomForest, predict_func = predict,
packages = "randomForest", fit_params = list(mtry = 4))

## $folds
##         repeat_1  repeat_2  repeat_3
## fold_1 0.9820183 0.9895682 0.8752296
## fold_2 0.8701763 0.8771651 1.2719188
## fold_3 1.1869986 0.7736392 1.3521407
## fold_4 1.0946892 1.1204090 0.7100938
## fold_5 0.9847612 1.0565001 0.9194678
## 
## $mean
## [1] 1.004318
## 
## $sd
## [1] 0.1791315
## 
## $median
## [1] 0.9847612

randomForest with parameter mtry = 4, and a validation set:

crossval::crossval_ml(x = X, y = y, k = 5, repeats = 2, p = 0.8,
fit_func = randomForest::randomForest, predict_func = predict,
packages = "randomForest", fit_params = list(mtry = 4))

## $folds
##                    repeat_1  repeat_2
## fold_training_1   1.0819863 0.9096807
## fold_validation_1 0.8413615 0.8415839
## fold_training_2   0.9507086 1.0014771
## fold_validation_2 0.5631285 0.6545253
## fold_training_3   0.7020669 0.9632402
## fold_validation_3 0.5090071 0.9129895
## fold_training_4   0.8932151 1.0315366
## fold_validation_4 0.8299454 0.7147867
## fold_training_5   0.9158418 1.1093461
## fold_validation_5 0.6438410 0.7644071
## 
## $mean_training
## [1] 0.9559099
## 
## $mean_validation
## [1] 0.7275576
## 
## $sd_training
## [1] 0.1151926
## 
## $sd_validation
## [1] 0.133119
## 
## $median_training
## [1] 0.9569744
## 
## $median_validation
## [1] 0.7395969

xgboost

In this case, the response and covariates are named ‘label’ and ‘data’. So (for now), we do this:

# xgboost example -----

require(xgboost)

f_xgboost <- function(x, y, ...) xgboost::xgboost(data = x, label = y, ...)

Fit xgboost with nrounds = 10:


crossval::crossval_ml(x = X, y = y, k = 5, repeats = 3,
  fit_func = f_xgboost, predict_func = predict,
   packages = "xgboost", fit_params = list(nrounds = 10,
   verbose = FALSE))

## $folds
##         repeat_1  repeat_2  repeat_3
## fold_1 0.9487191 1.2019850 0.9160024
## fold_2 0.9194731 0.8990731 1.2619773
## fold_3 1.2775092 0.7691470 1.3942022
## fold_4 1.1893053 1.1250443 0.7173760
## fold_5 1.1200368 1.1686622 0.9986680
## 
## $mean
## [1] 1.060479
## 
## $sd
## [1] 0.1965465
## 
## $median
## [1] 1.120037

Fit xgboost with `nrounds = 10, and validation set:

crossval::crossval_ml(x = X, y = y, k = 5, repeats = 2, p = 0.8,
  fit_func = f_xgboost, predict_func = predict,
   packages = "xgboost", fit_params = list(nrounds = 10,
   verbose = FALSE))

## $folds
##                    repeat_1  repeat_2
## fold_training_1   1.1063607 1.0350719
## fold_validation_1 0.7891655 1.0025217
## fold_training_2   1.0117042 1.1723135
## fold_validation_2 0.4325200 0.5050369
## fold_training_3   0.7074600 1.0101371
## fold_validation_3 0.1916094 0.9800865
## fold_training_4   0.9131272 1.2411424
## fold_validation_4 0.8998582 0.7521359
## fold_training_5   0.9462418 1.0543695
## fold_validation_5 0.5432650 0.6850912
## 
## $mean_training
## [1] 1.019793
## 
## $mean_validation
## [1] 0.678129
## 
## $sd_training
## [1] 0.147452
## 
## $sd_validation
## [1] 0.2600431
## 
## $median_training
## [1] 1.023388
## 
## $median_validation
## [1] 0.7186136

Note: I am currently looking for a gig. You can hire me on Malt or send me an email: thierry dot moudiki at pm dot me. I can do descriptive statistics, data preparation, feature engineering, model calibration, training and validation, and model outputs’ interpretation. I am fluent in Python, R, SQL, Microsoft Excel, Visual Basic (among others) and French. My résumé? Here!


Under License Creative Commons Attribution 4.0 International.

To leave a comment for the author, please follow the link and comment on their blog: T. Moudiki's Webpage - R.

R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.