Most analytics problems are path problems. The chart is the last inch. The miles are definition, ingestion, and the decision about which transformations are allowed to change the meaning of a field.
Skyneski draws the path as data, processing, models, intelligence, decisions. The last step is still a decision. Software can prepare it. Software should not be described as having made it when a person will be held to it.
Show the distribution
A single number is a poor picture of a racing record, a sensor history, or an operation. Comparison, change, and spread carry the information. Visualisation is doing its job when a reader can see what is typical and what is an exception without being told which conclusion to hold.
Predictive analytics belongs on this path when the estimate is labelled as an estimate. Historical analysis belongs on it when the past is allowed to be uneven. Both are poorer when they are collapsed into a ranking with no method.
Industry work, same standard
Race Monk is one place this standard meets a domain: horse racing, studied as data and history rather than as a tip. The same path applies to other domains. Specialisation changes the questions. It should not change the honesty of the diagram.