CityofToronto / bdit_volumes

Traffic Volumes Modelling Project
7 stars 6 forks source link

Investigate methods for assigning observations with missing variables to clusters #20

Open aharpalaniTO opened 7 years ago

aharpalaniTO commented 7 years ago

Ideally time-of-day related, but general methods may be OK as well. Simple and easy to understand method preferred.

aharpalaniTO commented 7 years ago

@sunnyqywang has done this. Calculation of sum(distance between profile and count for each 15-minute bin) for each cluster, with assignment based on lowest value. Will review effectiveness.

aharpalaniTO commented 7 years ago

For counts with less than 50% of data, hard to identify which cluster this location should belong to. To investigate using spatial properties as well (see #39)