The goal of the performance package is to provides a set of functions
to analyze and evaluate the performance of pairs of variables. It offers
a comprehensive set of performance metrics based on the variable types,
including continuous, categorical, and binary variables. The package
allows users to gain insights into the relationships between different
variables and assess their statistical significance.
You can install the development version of performance like so:
devtools::install_github("tpq/performance")In this example, we will demonstrate how to use the performance
package to analyze the performance of different variable types,
including continuous, categorical, and binary variables.
library(performance)Below we compare continuous measures.
# Create example data for continuous variables
var1_continuous <- c(1.2, 2.3, 3.1, 4.5, 5.7)
var2_continuous <- c(2.1, 3.5, 4.2, 5.3, 6.1)
# Compute performance measures for continuous variables
result_continuous <- performance(var1_continuous, var2_continuous)
# Print the results
print(result_continuous)
#> pearson_cor pearson_pval kendall_cor kendall_pval spearman_cor
#> cor 0.991135 0.001000623 1 0.01666667 1
#> spearman_pval mae mse
#> cor 0.01666667 0.88 0.852Below we compare binary measures.
# Create example data for binary variables
var1_binary <- c(0, 1, 1, 0, 1)
var2_binary <- c(1, 1, 0, 1, 0)
# Compute performance measures for binary variables
result_binary <- performance(var1_binary, var2_binary)
# Print the results
print(result_binary)
#> N X.tn. X.fp. X.fn. X.tp. acc spec X.sens. X.prec. X.f1. fpr
#> odds ratio 5 tn fp fn tp 0.2 0 sens prec f1 1
#> fnr fisher_or fisher_pval
#> odds ratio 0.6666667 0 0.4Below we compare categorical measures.
# Create example data for categorical variables
var1_category <- c("A", "B", "A", "C", "B", "A", "B", "C")
var2_category <- c("A", "C", "A", "C", "B", "A", "B", "C")
# Compute performance measures for categorical variables
result_category <- performance(var1_category, var2_category)
#> Warning in chisq.test(confusion): Chi-squared approximation may be incorrect
# Print the results
print(result_category)
#> acc chisq_stat chisq_pval
#> X-squared 0.875 11.55556 0.0209812Below we compare a mix of measures.
# Compute performance measures for mixed variables
result_mixed <- performance(var1_continuous, var2_binary)
# Print the results
print(result_mixed)
#> mean.0 mean.1 median.0 median.1 sd.0 sd.1 pct25.0 pct25.1 pct75.0
#> t 4.4 2.666667 4.4 2.3 1.838478 1.680278 3.75 1.75 5.05
#> pct75.1 ttest_t ttest_pval
#> t 3.4 1.068593 0.3926317Please note that this is a simplified example to demonstrate the usage of the package. In real-world scenarios, you can use the package to analyze more extensive and complex datasets, gain deeper insights, and make data-driven decisions based on the performance metrics.