{"Package":"ROCaggregator","Title":"Aggregate Multiple ROC Curves into One Global ROC","Version":"1.0.1","Authors@R":"\nperson(given = \"Pedro\",\nfamily = \"Mateus\",\nemail = \"pedro-cm@live.com.pt\",\nrole = c(\"aut\", \"cre\"),\ncomment = c(ORCID = \"0000-0002-3047-7630\"))","Description":"Aggregates multiple Receiver Operating Characteristic (ROC) curves\nobtained from different sources into one global ROC. Additionally, it<80><99>s\nalso possible to calculate the aggregated precision-recall (PR) curve.","License":"MIT + file LICENSE","Encoding":"UTF-8","RoxygenNote":"7.1.1","Imports":"utils, magrittr","Suggests":"testthat (>= 3.0.0), mockery, mockr, knitr, rmarkdown, ROCR,\npROC, pracma, stats","Config/testthat/edition":"3","VignetteBuilder":"knitr","URL":"https://gitlab.com/UM-CDS/general-tools/rocaggregator","BugReports":"https://gitlab.com/UM-CDS/general-tools/rocaggregator/-/issues","NeedsCompilation":"no","Packaged":"2021-08-09 15:29:58 UTC; admin","Author":"Pedro Mateus [aut, cre] ()","Maintainer":"Pedro Mateus ","Repository":"CRAN","Date/Publication":"2021-08-10 09:10:14 UTC","repoType":"cran","tarballUrl":"ftp://cran.r-project.org/pub/R/src/contrib/ROCaggregator_1.0.1.tar.gz","jsonAuthors":[{"name":"Pedro Mateus","email":"pedro-cm@live.com.pt","maintainer":true}],"readme":"\n\n\n# ROCaggregator\n\n\n\n\nAggregates multiple Receiver Operating Characteristic (ROC) curves\nobtained from different sources into one global ROC. Additionally, it’s\nalso possible to calculate the aggregated precision-recall (PR) curve.\n\n## Motivation\n\nThe ROC and the AUC (Area Under the Curve) can be essential metrics when\nevaluating a model. In situations where there is a parallelization of\nthe model development, such as federated learning, it becomes relevant\nto obtain precise measures for these metrics. These approaches usually\nproduce partial results that require aggregation methods to get the\ncomplete picture. The ROCaggregator appears in this context, allowing to\ncompute the precise ROC curve from the partial results.\n\n## Installation\n\nYou can install the released version of ROCaggregator from\n[CRAN](https://CRAN.R-project.org) with:\n\n``` r\ninstall.packages(\"ROCaggregator\")\n```\n\n## Example\n\nCheck the complete example provided in the vignette.\n\nObtain the global ROC curve from different sources by providing: - the\nfalse positive rate (fpr) - true positive rate (tpr) - thresholds\n(thresh) - the total number of negative samples/control - the total\nnumber of samples from each source\n\n library(ROCaggregator)\n\n y1 <- c(1, 0, 1, 1, 0, 0, 0, ...)\n # false positive rate values for each threshold\n fpr_1 <- c(0, 0, 0, 0, 0.002, ...)\n # true positive rate values for each threshold\n tpr_1 <- c(0, 0.004, 0.008, 0.012, 0.016, ...)\n # thresholds used\n thresh_1 <- c(0.9994038, 0.9986345, 0.99847864, 0.99575908, 0.99567612, ...)\n # count the number of negative labels\n negative_count_1 <- sum(y1 == 0)\n # total number of labels\n total_count_1 <- length(y1)\n ...\n # ROC curve\n roc <- roc_curve(\n list(fpr_1, fpr_2, ...),\n list(tpr_1, tpr_2, ...),\n list(thresh_1, thresh_2, ...),\n c(negative_count_1, negative_count_2, ...),\n c(total_count_1, total_count_2, ...)\n )\n\n # Precision-recall\n pre_recall <- precision_recall_curve(\n list(fpr_1, fpr_2, ...),\n list(tpr_1, tpr_2, ...),\n list(thresh_1, thresh_2, ...),\n c(negative_count_1, negative_count_2, ...),\n c(total_count_1, total_count_2, ...)\n )","jobInfo":{"package":"ROCaggregator","version":"1.0.1","parsingStatus":"success","parserVersion":1,"parsedAt":"2021-08-10T11:30:20+0000"}}

Description

Usage

Arguments

Value

Examples