In ml-e/ml4eda: Machine Learning for Experimental Data Analysis

knitr::opts_chunk$set(fig.width=12, fig.height=8, fig.path='Figs/',
                      warning=FALSE, message=FALSE)
options(width = 250)

library(knitr)
library(dplyr)

# results <- ml4edas::run_all()
results <- read.csv('summary_cv.csv')
results <- select(results, -X)
kable(results)

• Each row was aggregated from 1000 simulations.
• Prediction tests used 1000 permutations. I used mean squared error (on class probabilities). Absolute classification error did not give correctly sized tests, possibly due to ties.

ml-e/ml4eda documentation built on May 23, 2019, 2:03 a.m.