The result of "Request-Response" and "Batch" is different. Anyone know why? RRS feed

  • Question

  • I have deployed a web service from a ML experiment. It is a two-category classification problem. I found a very weird thing: the result of  "Request-Response" and "Batch" is different. I input the a group of feature in Request-Response interface and the scored label is "Yes". I clicked "csv" icon and use the generated csv to create an input file.  I uploaded the input file in the batch mode and run the test. However, the scored label is "No". I could not understand with same input why  "Request-Response" and "Batch" gave different scored label? Anyone has similar problem before?
    • Edited by MLLearner Wednesday, January 30, 2019 10:23 PM
    Wednesday, January 30, 2019 10:20 PM

All replies