Skip to content

What would be the best way to approach a classification problem with similar labels? #1

@adithyaan-creator

Description

@adithyaan-creator

Hi @srivatsan88 ,
I am working on a problem of text classification where the labels are quite similar, like

  • Bad reputation
  • customer issues
  • delays
  • good reputation

The thing is there is a major overlap between the first 3 labels, as many have common words and could fall into multiple categories.
Example - Delay could have words overlapping with bad reputation, same way with customer issues and bad reputation.
Is there any good approach to be taken that can ensure good metrics?
And what would be an ideal number of data points required. Currently there is only about 6000 data points.

Cheers.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions