search

insight-lane / crash-model

Build a crash prediction modeling application that leverages multiple data sources to generate a set of dynamic predictions we can use to identify potential trouble spots and direct timely safety interventions.
https://insightlane.org
MIT License
112 stars 40 forks source link

Update misleading documentation wording #221

Closed delewis13 closed 5 years ago

delewis13 commented 5 years ago

Under the "How does the project achieve its goal?" header:

"Predictions are generated on a per road-segment basis and will be made available via a searchable web visualization, with roads of highest risk easily identifiable. Details of which factors are most associated with risk on each road will also be included."

As far as I can tell, these factors are output by model.featuresimportances [for XGboost at least]. This gives feature importance for the data-set as a whole, rather than showing the feature weightings / contributions for a single road segment.

Should perhaps instead read "Details of which factors are most associated with risk for the dataset as a whole will also be included".

Cheers :)

bpben commented 5 years ago

Whoa, really delayed response on this, sorry. But you're right. We're actually working on explanatory features in #103 so until that's locked down we probably will leave as is. Thanks for the comment, though!