I am impressed by how quickly the notebook goes from raw data to interesting patterns. Nice.<p>If I wanted to be picky about statistics (I generally do), I'd note:<p>1. Equating enforcement-related statistics with usage-related statistics is a dangerous thing to do. For example:<p>a. The rate of non-CRACK incidents in TENDERLOIN is probably low (disproportionate with actual use) due to the focus limited police resources put on crack.<p>b. I would bet that the high rate of WEED incidents in INGLESIDE / BAYVIEW vs, say, MISSION, is due to police using weed enforcement as a weapon against gangs (and minorities). I mean, people smoke a lot of weed in the Mission.<p>2. Putting "AMPHETAMINE"-related crimes under "METH" is inaccurate, as is putting "OPIUM"/"OPIATE"-related crimes under "HALLUCINOGENIC".
This is an awesome exercise in data visualization. It's most striking to see how crack use was essentially flat until 2008, and then it started a steady decline. Looking at the charts, I would venture to interpret that the wave of gentrification following the market crash had the effect of disproportionately displacing crack users.<p>I wonder what this same study would look like across the bay, in Oakland.
It would be interesting to compare police data with a sewage study like this:
<a href="http://www.addictionjournal.org/press-releases/wastewater-analysis--a-new-source-of-monitoring-illicit-drug-use-in-europe" rel="nofollow">http://www.addictionjournal.org/press-releases/wastewater-an...</a><p>My guess would be that San Francisco's marijuana and cocaine consumption is at a per capita all-time high, even as police incidents decline.
<a href="http://lmart999.github.io/assets/drug_agg.png" rel="nofollow">http://lmart999.github.io/assets/drug_agg.png</a><p>Looking at this chart and being a bit assumptive, is it safe to say crack use has declined and meth has increased?<p>Maybe meth isn't really a new epidemic with a new user group, crack users just transferred to meth. Basically, those looking for cheap escapist drugs targeted at the very poor have moved onto a new product.<p>I'm curious which one is worse for society and the users?
Maybe I'm just obtuse, but this graph makes no sense to me: <a href="http://lmart999.github.io/assets/cluster_gis.png" rel="nofollow">http://lmart999.github.io/assets/cluster_gis.png</a><p>What exactly do the colors mean? And what do those brackets at the top and at the side mean?
Are there any resources on how to properly format raw data t make it most useful for display on a map? I'm hoping to do a crime map for my city and haven't quite found a good place to start.