<IWNMWEIWNMWE[m]>
Hey guys as you know i was planning to put in a proposal for implementation of xgboost but someone has already given a proposal in the mailing list for it with some common objectives hence I would like to know about what would happen in such a case and if should stick to this idea or propose a different one
Emiya[m] has joined #mlpack
MahendraChaudhar has joined #mlpack
robobub has quit [Quit: Connection closed for inactivity]
<akhunti1[m]>
If u see res is the dest data i gave using url command like this : curl -X POST -H 'Content-Type: application/json' -d '{"data": { "ndarray": [[1,2,3,4,5,6]]}}' http://localhost:9000/api/v1.0/predictions
<akhunti1[m]>
ndarry is a dest data
<akhunti1[m]>
but the accuracy mean prob i got 12 , expected is i should get 2 probability .
<akhunti1[m]>
* got 12 point , expected
<rcurtin[m]>
looks like the first two elements of the accuracy are the probability of the first class and the probability of the second class (i.e. the probability that it is not the first class :)); you can see that they sum to 1
<rcurtin[m]>
IWNMWE (IWNMWE): that's okay, multiple people can submit applications for one project
<akhunti1[m]>
rcurtin: Then why other 10 probality got created . because when i am giving same data point to cli it is giving me 2 probality . and both sum is 1 .
<rcurtin[m]>
oh, I see what you mean now, sorry; what is the shape of the input matrix? mlpack's FFN will expect one point to correspond to one column (so it should have size 6 rows x 1 column), but if you gave that as 1 row x 6 columns, then the FFN will make predictions for 6 points, not one
<rcurtin[m]>
if the shape is actually 1 row by 6 columns, just use .t() to transpose it to 6 rows by 1 column 👍️
<rcurtin[m]>
personally I don't have time to go through the code, you'll have to investigate the details, but the basic idea of what I wrote above should be correct
<akhunti1[m]>
did u mean rf.Classify(arma::mat(res).t(), testPredictions,probs); [ if the shape is actually 1 row by 6 columns ]
<rcurtin[m]>
yeah, that looks about right
VanshSharma[m] has quit [Quit: You have been kicked for being idle]
<akhunti1[m]>
Hi sorry , but this way : rf.Classify(arma::mat(res).t(), testPredictions,probs); i am not able to transpose mat .
<akhunti1[m]>
I also tried rf.Classify(arma::mat(res).st(), testPredictions,probs); .st() method , but did not work .
robobub has joined #mlpack
<vaibhavp[m]>
<akhunti1[m]> "Hi sorry , but this way : rf...." <- What error are you getting?