In jbryer/mldash: Evaluating Machine Learning Models Across Many Datasets

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(mldash)

rf_model <- mldash::new_model(
    name = 'randomForest_classification',
    type = 'classification',
    description = 'Random forest prediction model usign the randomForest R package.',
    train_fun = function(formula, data) {
        randomForest::randomForest(formula = formula, data = data, ntree = 1000)
    },
    predict_fun = function(model, newdata) {
        randomForest:::predict.randomForest(model, newdata = newdata, type = "prob")[,2,drop=TRUE]
    },
    packages = "randomForest",
    overwrite = TRUE
)

Results in the following file:

name: randomForest_classification
type: classification
description: Random forest prediction model usign the randomForest R package.
train: function (formula, data)
    {
        randomForest::randomForest(formula = formula, data = data,
            ntree = 1000)
    }
predict: function (model, newdata)
    {
        randomForest:::predict.randomForest(model, newdata = newdata, type = "prob")[,2,drop=TRUE]
    }
packages: randomForest
note:

Note that for classification models, the run_models() function will ensure that the dependent variable is coded as a factor. If the model assumes another data type (e.g. TRUE or FALSE) it will need to convert the variable. Otherwise, the data files (read in by the read_data() function) should ensure all independent variables a properly coded.

jbryer/mldash documentation built on March 4, 2023, 9:35 p.m.