Figure 2
From: Mixture models for analysis of melting temperature data
AIC plotted for models with various numbers of categories. This AIC plot corresponds to 100 data points generated from four temperatures with 5 × σ separations (80, 80.3, 80.6, 80.9). The minimum at four indicates that a model with four categories is optimal. Models with 1–3 categories have a high AIC, indicating that they do not sufficiently explain the data. Models with 5–7 categories do not improve the correlation of the model to the data points sufficiently to justify the additional parameters estimated.