I am really sorry if it is trivial. Why is that running the trainlogistic with different rates and passes give me same auc and confusion matrix result on the test data. Although the time of training increases with number of passes but seems to have no effect on result. I can understand that rate is the learning rate but what is the purpose of passes?
I have split my data into 80% training and 20% Test data. Training data size is 51000 (approx) Test data size is 12000(approx) Total size of the data is 63000 (approx) I have gone through the explanations in MIA book chapter 13 but is there any rule of thumb for deciding it? Also, while training the model , the following output is not intuitive to comprehend especially the last part of the output having quite a few zeros. Any thoughts? CLASS ~ -13.774*AON + 10.065*BALANCE + 28804.244*INDECRE + -0.782*INDECRE_FREQ + -634.428*Intercept Term + 28804.244*MOU + -1312.785*NO_VOICE_CALLS + -24613.959*OFFNET_USAGE + 118998.287*ONNET_USAGE + -634.428*RECHARGE + 118998.287*RECHARGE_FREQ + -1312.785*SMS AON -13.77429 BALANCE 10.06540 INDECRE 28804.24420 INDECRE_FREQ -0.78217 Intercept Term -634.42839 MOU 28804.24420 NO_VOICE_CALLS -1312.78468 OFFNET_USAGE -24613.95945 ONNET_USAGE 118998.28731 RECHARGE -634.42839 RECHARGE_FREQ 118998.28731 SMS -1312.78468 -24613.959449080 28804.244197092 0.000000000 0.000000000 -1312.784681776 -634.428388766 10.065402474 0.000000000 -0.782170605 -13.774291103 118998.287306148