roc.plot.calculate calculates PCC, sensitivity, specificity, and Kappa for a single presence absence model at a series of thresholds in preparation for creating a ROC plot.
roc.plot.calculate(DATA, threshold = 101, which.model = 1, na.rm = FALSE)Returns a dataframe where:
| [,1] | threshold | thresholds used for each row in the table |
| [,2] | PCC | percent correctly classified |
| [,3] | sensitivity | |
| [,4] | specificity | |
| [,5] | Kappa |
a matrix or dataframe of observed and predicted values where each row represents one plot and where columns are:
DATA[,1] | plot ID | text | |||
DATA[,2] | observed values | zero-one values | |||
DATA[,3] | predicted probabilities from first model | numeric (between 0 and 1) | |||
DATA[,4] | predicted probabilities from second model, etc... |
cutoff values between zero and one used for translating predicted probabilities into 0 /1 values, defaults to 0.5. It can be a single value between zero and one, a vector of values between zero and one, or a positive integer representing the number of evenly spaced thresholds to calculate.
a number indicating which model from DATA should be used
a logical indicating whether missing values should be removed
Elizabeth Freeman eafreeman@fs.fed.us
roc.plot.calculate is a streamlined version of presence.absence.accuracy designed specifically to compute the accuracy measures needed to produce a ROC plot. roc.plot.calculate is less versatile, but more efficient than presence.absence.accuracy.
Unlike presence.absence.accuracy, roc.plot.calculate will only work for a single set of model predictions. Therefore either DATA can contain only one model prediction, or which.model must be used to indicate a single model prediction from DATA. By default, if DATA contains more than one model prediction, and which.model is not specified, roc.plot.calculate will use the first model prediction (e.g. DATA[,3]).
roc.plot.calculate was written as a sub-function for the plotting functions(i.e. error.threshold.plot, auc.roc.plot, but can be used on its own to produce a simple table of how the accuracy measures vary with choice of threshold.
To produce attractive plots requires a large number of thresholds. The default value of threshold = 101 is a good compromise between speed and resolution.
cmx, pcc, sensitivity, specificity, Kappa, presence.absence.accuracy, error.threshold.plot, auc.roc.plot
data(SIM3DATA)
roc.plot.calculate(SIM3DATA,which.model=2)
Run the code above in your browser using DataLab