Activist Data Mining Solutions

1. No pivot design for the given inputs.
 
num,prenitro,precarbon,prelight,postnitro,light,carbon
1,0,0,1,0,1,1
2,0,1,0,1,0,1
3,1,1,0,0,0,0
4,1,0,1,1,1,0
5,0,0,1,2,0,0
6,1,1,0,2,1,1
7,1,0,0,3,0,1
8,0,1,1,3,1,0

2a. For a prelight pivot, add the following 8 (take the prelight
values in the no pivot and flip their values to get the 8
additional experiments that will give minimal pairs):
p1,0,0,0,0,1,1
p2,0,1,1,1,0,1
p3,1,1,1,0,0,0
p4,1,0,0,1,1,0
p5,0,0,0,2,0,0
p6,1,1,1,2,1,1
p7,1,0,1,3,0,1
p8,0,1,0,3,1,0

2b. For a prelight half-pivot, need only add 
an experiment having all 0s and
the following four
(all the ones that had 0 prelight in the no pivot should change to light):
hp1,0,1,1,1,0,1
hp2,1,1,1,0,0,0
hp3,1,1,1,2,1,1
hp4,1,0,1,3,0,1

3. The key point to note is that the threshold for a node in the
shifted wavelet tree must be low enough to capture any interval
not caught by lower nodes in the tree. 

node | alarm thresh
2 | 4
4 | 4
8 | 4 (to cover the case where a window of size 5 has 4 events)
16 | 5
32 | 7

4a. Given a schema
ticks(ID, date, endofdayprice)
find the maximum and minimum return of each stock
where return is the ratio of the price of day i compared with 
the price of day i-1.

select ID, max (price/prev(price)), min (price/prev(price))
from ticks
assuming ORDER date
group by ID

4b. Assuming ticks were already sorted by ID, how would
you process the above query if you had complete control
of the optimization process?

Sort within each ID by date, compute price/prev(price) and
on the fly compute the minimum and maximum.