correlated {tiger} | R Documentation |
Calculate the correlation structure between multiple performance measures
correlated(result, limit = 0.85, plot.scatter = FALSE, keep = NA) correl(measures, limit = 0.85, plot.scatter = FALSE, keep = NA)
result |
object returned from tiger |
measures |
data.frame for which to determine correlation structure |
limit |
Limit for absolute correlation, above which data is considered to be correlated |
plot.scatter |
Boolean, indicating whether to show pairwise plots for correlated measures |
keep |
Vector with names of measures that must not be excluded because of correlation with other measures |
correl returns:
pairs |
Matrix with indices of pairwise correlated measures |
pairs.by.name |
Matrix with measure names of pairwise correlated measures |
possible.exclusion |
List indicating which measures might be removed to end up with no strongly correlated measures. The list also indicates, which measure is correlated to the removed measures |
to.drop |
List of indices for measures to drop (according to previous list) |
to.drop.by.name |
List of measure names (of the previous list) |
correlated returns a list of two correl results, one for the
original performance measures and one for the transformed measures
from a result from tiger
.
Dominik Reusser
This method helps to reduce the amount of data to be analyzed from an evaluation using tiger
data(tiger.example) correlated <- correlated(tiger.single, keep=c("CE","RMSE" ))