search

thekingofkings / chicago-crime

Crime correlation anaysis
MIT License
11 stars 3 forks source link

Explore the properties of various crime categories #18

Closed thekingofkings closed 7 years ago

thekingofkings commented 7 years ago

The correlation between urban features and various crime categories are different.

We observe that in Theft crime category, the POI features can improves the prediction accuracy, while the Geo features are useless. On the other hand, for Motor vehicle Theft and Robbery, the POI features make the prediction worse.

How to explain these observations?

Data available

For crime data, we have more detailed information such as date / time, general description. Previously, we did not use any of those data.

Things to try

Some crime categories behave differently from others. We can try to define / measure the similarity of two crimes (with the data we have). Although it is hard to explain why some urban features has such correlation, but it is helpful to show which crime property makes the correlations are different.

thekingofkings commented 7 years ago

Code base change

Add new function to CrimeDataset class inside the python/Crime.py.

def temporalDistribution_perTract_perCategory()

This function takes one-year raw crime data (crime-file-name) as input, and outputs the occurrence time distribution of various crimes by categories for each tract area.