Validating a global urbanisation model

For this study, predictive modelling is used to validate a high-resolution model that simulates urban development. A logit model is used to predict the likelihood of a cell being classified as urban or non- urban based on historical data. Several essential choices had to be made during the validation process about the classification model and the model evaluation metrics. This presentation focuses on three important choices in performing a validation analysis: choosing between a static or dynamic analysis, selecting an appropriate performance indicator, and deciding between a linear or non-linear classification.