This week's task is to take our model to data for the purpose of learning something about the data. We have worked hard to build a causal model which we then used to generate synthetic data for two customer bases. In this episode we introduce ourselves to an hypothetical-deductive principled approach to probabilistic inference. It is Bayesian in nature, but in the end it simply indicates the plausibility of a range of mutually exclusive - completely exhaustive (MECE) hypotheses (our causal model for instance) consistent with sampled data we collected. We will use this model with various observed (samples) and unobserved (hypotheses) data to help inform us of tradeoffs in decisions.
Course site: systemdynamics...
23 окт 2024