watson_segment: Watson Advises Sherlock By Detecting Segments and Assigning...
In Netflix/sherlock: Causal Machine Learning for Segment Discovery and Analysis

Dr. James H. Watson was Sherlock's friend, flatmate, and assistant. He often contributed to Sherlock's work, usually by accompanying him in the field and providing advice that could guide Sherlock towards a solution to the case. Within sherlock's causal segmentation framework, watson_segment mirrors Dr. Watson's role by assigning a treatment rule after Sherlock has completed a preliminary consultation via sherlock_calculate. The output is a slightly augmented version of the data structure produced by sherlock_calculate; however, this step is necessary. After Dr. Watson has helped Sherlock in finalizing the segmentation analysis, the segmentation quality may be evaluated via mycroft_assess.

1 2	watson_segment(data_with_consult, segment_fun, ..., type = c("inferential", "analytic"))

`data_with_consult`	A `data.table` containing the output `sherlock_calculate`, which matches the input data, augmented with cross-validated nuisance parameter estimates and an estimate of the CATE. A summary of the estimates across segmentation strata is made available as an attribute of the `data.table`.
`segment_fun`	A particular choice of function for the assignment of a treatment rule to a segment based on a specified threshold or constraint. For details on these, consult the documentation in `cost_funs`.
`...`	Additional arguments passed to `segment_fun`. For details, see the documentation for the cost functions in `cost_funs`.
`type`	A `character` string (of length one) specifying how the treatment decision based on the CATE is to be made. There are two options: `"inferential"` (the default) uses a hypothesis test to evaluate whether the estimated CATE is statistically different from a threshold and assigns a treatment decision based on the resultant p-value. `"analytic"` simply evaluates whether the estimated CATE exceeds a given threshold and assigns treatment to segments for which this holds. Note that in both cases the threshold must be either provided directly (as an input to `cost_threshold`) or will be discovered based on specified constraints (as in `cost_budget`).