Skip to main content

Advertisement

Figure 1 | Population Health Metrics

Figure 1

From: Direct estimation of cause-specific mortality fractions from verbal autopsies: multisite validation study using clinical diagnostic gold standards

Figure 1

The process of generating 500 test and train datasets and applying KL estimation to them. After dividing the whole dataset into 25% testing and 75% training portions (randomly, stratified by cause), a draw from an uninformative Dirichlet distribution was used to perturb the cause combination of the test set (by resampling each cause with replacement according to a CSMF that was drawn from Dirichlet distribution). Accuracy of the KL method was calculated by comparing the KL-estimated CSMFs and the true CSMF of the test dataset.

Back to article page