#--------------------------------------------------------------------------------------------------
#--------------------------------------------------------------------------------------------------
getMetaLevelData = function(all.files) {
aux = lapply(all.files, function(file){
# print(file)
values = load(file, verbose = FALSE)
ret = get(values)
return(ret)
})
df = do.call("rbind", aux)
# selecting only 95% of the statistical rule
df = df[grepl(df$task, pattern = "_95_"), ]
# featsel == none ,tuning == none, balance == none,norm == no_norm, resamp == 10-CV
sub = dplyr::filter(df, featsel == "none" & tuning == "none" & balance == "none"
& norm == "no_norm" & resamp == "10-CV")
# selecting just the simple results
# [task, algo, auc, rep]
select.columns = c("task", "algo", "auc", "rep")
ret = sub[,which(colnames(sub) %in% select.columns)]
return(ret)
}
#--------------------------------------------------------------------------------------------------
#--------------------------------------------------------------------------------------------------
aggregateMetaLevelData = function(meta.data) {
algos = unique(meta.data$algo)
tasks = unique(meta.data$task)
combinations = expand.grid(algos, tasks)
aux = lapply(1:nrow(combinations), function(k) {
sel = dplyr::filter(meta.data, task == combinations[k,2] & algo == combinations[k,1])
sel = sel[1:30, ]
ret = sel[1,1:2]
ret$avgAuc = mean(sel$auc)
ret$sdAuc = sd(sel$auc)
return(ret)
})
avg.df = do.call("rbind", aux)
avg.df$Setup = gsub(x = avg.df$task,
pattern = "classif.J48_95_165d_|classif.rpart_95_165d_|_original_dist|classif.J48_165d_95_|classif.rpart_165d_95_",
replacement = "")
# renaming task
avg.df$task = gsub(x = avg.df$task, pattern = "classif.|_95_165d_|_165d_95_|complex_|simple_|all_|original_dist",
replacement = "")
avg.df$task = as.factor(avg.df$task)
avg.df$algo = factor(avg.df$algo, levels = c("RF", "SVM", "GP", "KNN", "CART", "NB", "LR"))
return(avg.df)
}
#--------------------------------------------------------------------------------------------------
#--------------------------------------------------------------------------------------------------
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.