Calculate True Positives in R

What Are True Positives?

In binary classification problems, true positives (TP) refer to cases where the model correctly identifies positive instances. They represent the number of items that are both actually positive and correctly classified as positive by the model.

True positives are one of the four possible outcomes in a confusion matrix:

• True Positive (TP): Correctly identified positive cases
• False Positive (FP): Incorrectly identified positive cases
• True Negative (TN): Correctly identified negative cases
• False Negative (FN): Incorrectly identified negative cases

How to Calculate True Positives

The calculation of true positives depends on the context of your classification problem. In the simplest case, you can count the number of positive predictions that match the actual positive cases.

In machine learning, you typically calculate true positives by comparing the predicted labels with the actual labels in your test set.

Example Calculation

Suppose you have a dataset of 100 patients, where 30 actually have a disease (positive cases) and 70 do not. Your model predicts 25 patients as positive. If 20 of these predictions are correct, then:

R Implementation

In R, you can calculate true positives using the caret package or by manually comparing predicted and actual values. Here's an example using the caret package:

library(caret)
# Example data
actual <- c(rep(1, 30), rep(0, 70))  # 30 positives, 70 negatives
predicted <- c(rep(1, 25), rep(0, 75))  # 25 predicted positives

# Create confusion matrix
confusionMatrix <- confusionMatrix(factor(predicted), factor(actual))
true_positives <- confusionMatrix$table[2,2]
print(paste("True Positives:", true_positives))

This code will output the number of true positives in your dataset.

Interpretation of Results

The number of true positives alone doesn't tell the whole story. You should also consider:

• False positives (Type I errors)
• False negatives (Type II errors)
• Precision (TP / (TP + FP))
• Recall (TP / (TP + FN))
• F1 score (harmonic mean of precision and recall)

A high number of true positives is good, but it should be balanced with other metrics to ensure your model is performing well overall.