This R package ports the following metrics from the scikit-learn machine learning package for Python:
| Function | Description |
|---|---|
trapz |
Computes the area under a curve using the trapezoidal rule |
precision_recall_curve |
Computes precision, recall, and threshold values |
roc_curve |
Computes true positive rates, false positive rates, and threshold values |
average_precision_score |
Computes the area under the precision-recall curve |
roc_auc_score |
Computes the area under the Receiver Operating Characteristic (ROC) curve |
I've attempted a faithful port including most unit tests and error checking. Comments and variable names are maintained where possible, roughly mapping to Hadley's style guide.
To install:
git clone https://github.com/shwhalen/fastmetrics
R CMD INSTALL fastmetrics
To use:
library(fastmetrics)
set.seed(0)
labels <- c(rep(0, 50), rep(1, 50))
predictions <- runif(100)
roc_auc_score(labels, predictions)
[1] 0.4624
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent