For this project I wanted to be able to further research my look into finding outliers through comparison of normalized averages and standard deviations

SELECT hour(day1.cdate),

sum(case when deweyClass>=810 and deweyClass<820 and dayofweek(day1.cdate) = 1 then 1 else 0 end),

sum(case when deweyClass>=810 and deweyClass<820 and dayofweek(day1.cdate) = 2 then 1 else 0 end),

sum(case when deweyClass>=810 and deweyClass<820 and dayofweek(day1.cdate) = 3 then 1 else 0 end),

sum(case when deweyClass>=810 and deweyClass<820 and dayofweek(day1.cdate) = 4 then 1 else 0 end),

sum(case when deweyClass>=810 and deweyClass<820 and dayofweek(day1.cdate) = 5 then 1 else 0 end),

sum(case when deweyClass>=810 and deweyClass<820 and dayofweek(day1.cdate) = 6 then 1 else 0 end),

sum(case when deweyClass>=810 and deweyClass<820 and dayofweek(day1.cdate) = 7 then 1 else 0 end)

from (SELECT cout as cdate, deweyClass FROM ( SELECT cout, deweyClass FROM spl2.inraw WHERE DATE(cout) >= '2013-03-01' AND DATE(cout) < '2013-04-01' ) as outday1 where hour(outday1.cout) >=10 and hour(outday1.cout)<= 20 ) as day1 group by hour(day1.cdate) order by hour(day1.cdate)

sum(case when deweyClass>=810 and deweyClass<820 and dayofweek(day1.cdate) = 1 then 1 else 0 end),

sum(case when deweyClass>=810 and deweyClass<820 and dayofweek(day1.cdate) = 2 then 1 else 0 end),

sum(case when deweyClass>=810 and deweyClass<820 and dayofweek(day1.cdate) = 3 then 1 else 0 end),

sum(case when deweyClass>=810 and deweyClass<820 and dayofweek(day1.cdate) = 4 then 1 else 0 end),

sum(case when deweyClass>=810 and deweyClass<820 and dayofweek(day1.cdate) = 5 then 1 else 0 end),

sum(case when deweyClass>=810 and deweyClass<820 and dayofweek(day1.cdate) = 6 then 1 else 0 end),

sum(case when deweyClass>=810 and deweyClass<820 and dayofweek(day1.cdate) = 7 then 1 else 0 end)

from (SELECT cout as cdate, deweyClass FROM ( SELECT cout, deweyClass FROM spl2.inraw WHERE DATE(cout) >= '2013-03-01' AND DATE(cout) < '2013-04-01' ) as outday1 where hour(outday1.cout) >=10 and hour(outday1.cout)<= 20 ) as day1 group by hour(day1.cdate) order by hour(day1.cdate)

I found that by using a color pallete where the insignificant data was the same color as the background, it became much easier to find the outliers. I also found that having the seperate modes helped significantly in understanding the data.

chronological

sorted by average

sorted by standard deviation

chronological

sorted by average

sorted by standard deviation

I was able to find a better color scheme that would look more similar to a heat mapping however I was having an issue in that it was hard to tell the relevance of any of the color. What's dense? what's sparse? everything seemed equally important and this simply was not the case.

sorted by average

sorted by standard deviation

sorted by standard deviation

I used Processing.

Control: A = sort by average

C = chronological

S = sort by standard deviation

Source Code + Data

Control: A = sort by average

C = chronological

S = sort by standard deviation

Source Code + Data