Open coryschillaci opened 9 years ago
I can also get more mturk data (re - law of large numbers :)
Let's merge what we have... My personal assessment is also already very biased by the mturk data analysis.
P
More turk data won't hurt if there's money to pay for it! If you do run, can you put a test that they correctly identify most of the moodids that occur as circumplex affects as well in order to get credit? The ones that occur in both lists are:
I checked the idea of scoring the MT responders by percent of the above list that they assigned correctly to the identical circumplex affect. The majority vote gets 100%, which is good. It looks like the first five do much better than the last five.
yeah, so we need more day data... i will not post then the job till tomorrow
thanks
p
On Wednesday, March 11, 2015, coryschillaci notifications@github.com wrote:
FYI I checked the idea of scoring the MT responders by percent of the above list that they assigned correctly to the identical circumplex affect. The majority vote gets 100%, which is good. It looks like the first five do much better than the last five.
— Reply to this email directly or view it on GitHub.< https://ci6.googleusercontent.com/proxy/Rvffjd-RDYE66Qhd3XqWI38TS3UGQcDIlCu_EcYJHLCV_lxK8J5LxGqnw7ssBpGkWI7xMp755bYWfV2eRZ2bXUQjkg5bGFF0z7UtvPsX-dwmHrGn-bvgQpOtiK1Xtc0c6UW1Qc6rruIL5K0jVp9-MdQ_hzsW9g=s0-d-e1-ft#https://github.com/notifications/beacon/AAxWcS70p-hXZXawKVeyhdrVMW1KOFuVks5n0RRNgaJpZM4DtFWf.gif
@anasrferreira and @peparedes Should also upload their manual responses, then we need a strategy to merge with the Mech Turk data.