R中逻辑回归模型的ConfusionMatrix

Question

我正在尝试在我的逻辑回归模型上计算混淆矩阵，但出现错误

错误：数据和参考应该是相同水平的因素。 另外：警告消息：在 is.factor(reference) 中：引入了 NA 强制

代码：

usage_model_rl <- glm(data = train_usage_rl, formula = Subscription ~., family = binomial(link = "logit"))
usage_predictions <- predict(usage_model_rl, test_usage_rl, type = "response") %>%
  bind_cols(test_usage_rl %>% dplyr::select(Subscription), preds = .)


cut04 <- usage_predictions %>%mutate(predicted = ifelse(preds >= 0.4, 1, 0)) %>% dplyr::select(-preds) %>% dplyr::select(predicted, Subscription) %>%mutate_all(list(~ factor(., levels = c(1, 0)))) %>%table()

confusionMatrix(usage_predictions,as.numeric(test_usage$Subscription),positive==1)

问候

Answer 1

试试这个：

confusionMatrix(usage_predictions,test_usage$Subscription,positive="1", dnn=c("Prediction", "Reference"))

确保它们都是因素。