{"items": [{"author": "BDan", "source_link": "https://www.facebook.com/jefftk/posts/286002998078798?comment_id=286051804740584", "anchor": "fb-286051804740584", "service": "fb", "text": "The temperatures in your redefined scale should actually be -80J and -90J; it looks to me like what you really mean is that adding squared temperatures doesn't work when some of the temperatures are negative, which seems obvious. So it should work fine if you do everything in Kelvin and set your constants appropriately, or, as in the lilacs' case, define your conditions such that you never have to deal with negative numbers.", "timestamp": "1317835162"}, {"author": "Jeff Kaufman", "source_link": "https://www.facebook.com/jefftk/posts/286002998078798?comment_id=286072828071815", "anchor": "fb-286072828071815", "service": "fb", "text": "@BDan: I've fixed the typo.
When adding squares, it matters where your zero point is. This criteria is not the same when expressed in kelvin. Looking at five 10C days we see:
sqrt(sum(10C*10C for i in range(5))) = 22C (295K)
sqrt(sum((283.15K)*(283.15K) for i in range(5))) = 633K (360C)
If we do this in raw kelvins we decide it's plenty warm enough to bloom but if we do it in raw centigrade we decide it's not.
The answer is that we should be squaring temperature differences. If we did this all using the second formulation (with t-t_freezing) it would all work.", "timestamp": "1317837733"}, {"author": "David Chudzicki", "source_link": "https://plus.google.com/106120852580068301475", "anchor": "gp-1317857334220", "service": "gp", "text": "I want data. ", "timestamp": 1317857334}, {"author": "Jeff Kaufman", "source_link": "https://plus.google.com/103013777355236494008", "anchor": "gp-1317865143063", "service": "gp", "text": "TSOR doesn't turn anything up. I would think there would be some nice free archive of average daily temperatures for machine learning purposes.", "timestamp": 1317865143}, {"author": "David Chudzicki", "source_link": "https://plus.google.com/106120852580068301475", "anchor": "gp-1317865224097", "service": "gp", "text": "how about when the lilacs bloom?", "timestamp": 1317865224}, {"author": "Jeff Kaufman", "source_link": "https://plus.google.com/103013777355236494008", "anchor": "gp-1317865951266", "service": "gp", "text": "Data collected carefully over six years in many locations between 1842 and 1848 isn't good enough for you?", "timestamp": 1317865951}, {"author": "David Chudzicki", "source_link": "https://plus.google.com/106120852580068301475", "anchor": "gp-1317866131690", "service": "gp", "text": "I must have missed that. It's in one of your references?", "timestamp": 1317866131}, {"author": "Jeff Kaufman", "source_link": "https://plus.google.com/103013777355236494008", "anchor": "gp-1317870156228", "service": "gp", "text": "I read it in one of the older ones, but can't find it now. Anyway, I was being silly: what we really want is the raw data for temperature and blooming, and I don't see that anywhere.", "timestamp": 1317870156}, {"author": "David Chudzicki", "source_link": "https://plus.google.com/106120852580068301475", "anchor": "gp-1317870321767", "service": "gp", "text": "yeah", "timestamp": 1317870321}]}