| Type: | Package |
| Title: | Beta Calibration |
| Version: | 0.1.0 |
| Author: | Telmo M Silva Filho and Meelis Kull |
| Maintainer: | Telmo M Silva Filho <tmfilho@gmail.com> |
| Description: | Fit beta calibration models and obtain calibrated probabilities from them. |
| License: | MIT + file LICENSE |
| Encoding: | UTF-8 |
| LazyData: | true |
| NeedsCompilation: | no |
| Packaged: | 2017-02-24 22:46:47 UTC; tmfilho |
| Repository: | CRAN |
| Date/Publication: | 2017-02-25 01:03:32 |
Beta Calibration
Description
Builds a beta calibration model on probability vector p and label vector y, fitting the parameters chosen by the user, with possible values being "abm", "ab" and "am". Returns the calibration model, the calibration map and the chosen parameters.
Usage
beta_calibration(p, y, parameters="abm")
Arguments
p |
A vector of probabilities that will be used to train the calibration model. |
y |
A vector of labels that will be used to train the calibration model. |
parameters |
The parameters that will be fitted by the model. |
See Also
Examples
## Creating a vector of probabilities
p <- seq(0.01,0.99,0.01)
## Creating a label vector based on the probability vector
y <- rbinom(99,1,p)
## Fitting beta calibration with three parameters
calib <- beta_calibration(p, y, "abm")
## Fitting beta calibration with two shape parameters
calib <- beta_calibration(p, y, "ab")
## Fitting beta calibration with one shape parameter and one location parameter
calib <- beta_calibration(p, y, "am")
Predict Calibrated Probabilities
Description
Returns calibrated probabilities from calib$model, where calib is obtained by calling the beta_calibration function.
Usage
beta_predict(p, calib)
Arguments
p |
A vector of probabilities that the model will calibrate. |
calib |
A list containing a calibration map, a calibration model and the fitted parameters, obtained by calling the |
See Also
Examples
## Creating a vector of probabilities
p <- seq(0.01,0.99,0.01)
## Creating a label vector based on the probability vector
y <- rbinom(99,1,p)
## Fitting beta calibration with three parameters
calib <- beta_calibration(p, y, "abm")
## Obtaining calibrated probabilities
probas <- beta_predict(p, calib)