Skip to content

Training and test data #1

@cemevin

Description

@cemevin

Hi,

I checked your code and noticed that you divide training and test data randomly, based on a split ratio. However in the case of stock markets, I dont think you can do that, since this data is time sensitive. Future prices may imply past prices. When I edited your code to reflect this change (divide the first 70% of the data as training set and last 30% as test set), the accuracy drops down to 50%, unfortunately. I wish it didnt :)

Best,
Cem

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