Results can be directly accessed here.

• create a way to score results, i.e. can we assess how well the algorithm identified a known serial criminal's residence / place of work
• allow for "confidence" of location, e.g. some of the disappearance locations are "last seen in the Seattle area" (Patricia Yellowrobe)
• compensate score values for water, unreasonableness of commute or excessive commute times
  • Rossmo discusses this on pg. 154 - 155

    With no potential victims situated in the English Channel to the south, he [Jose Rodrigues] was forced to confine his attacks to locations north of his residence, which resulted in a distorted target pattern

• move calculations into a distinct repo / project, separate from a specific implementation (this repo)
• adjust size of plot points; right now, their display leaves a bit to be desired (they're too small when zoomed out). Bokeh seems limited in its ability to support dynamic glyph sizes based on Zoom.
• address version conflicts with nbclient and jupyter_client; it's been super finnicky to get the book to consistently update without errors
• varying values for f and g are not readily supported; in many cases, users will run into RuntimeWarning: invalid value encountered in double_scalars due to infinitesimally small values
• cleanup / add more type hinting