Figure 4. Confusion matrix depicting the prediction accuracies and inaccuracies for eight demographic models using delimitR for model selection, which involves the simulated mSFS and ‘abcrf’. Rows indicate the model the data were simulated under and columns indicate the model that was predicted, each cell then indicates the proportion of simulated data under the true model that is classified as the predicted model. Thus, the diagonal cells of the matrix depict the proportion of correct model classifications, as the predicted model aligns with the true model, whereas the off-diagonal cells depict the proportion of model simulations that are incorrectly classified, and specifically which model the simulations are incorrectly classified as.