Creates a model that assumes events are equally likely across all regions. This serves as a "null hypothesis" baseline - regions where events cluster will show high surprise under this model.
Details
Under the uniform model, expected probability for each region is 1/n, where n is the total number of regions. The likelihood is computed as:
$$P(D|Uniform) = 1 - \frac{1}{2} \sum_i |O_i - \frac{1}{n}|$$
This is the Total Variation Distance from uniform, transformed to a probability.
The uniform model is useful for detecting spatial clustering - any concentration of events in fewer regions will produce high surprise.
Examples
# Create uniform model
model <- bs_model_uniform()
# The model computes likelihood when used in a model space
space <- model_space(model)