Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add Parquet Format to Store Nomalized Data #131

Closed
zhangpengshan opened this issue Jun 1, 2015 · 4 comments
Closed

Add Parquet Format to Store Nomalized Data #131

zhangpengshan opened this issue Jun 1, 2015 · 4 comments
Assignees
Milestone

Comments

@zhangpengshan
Copy link
Contributor

In Shifu, nomarlied data needs some space to store. To save space and improve performance, try ParquetStorer and ParquetInputFormat in normalize.pig and trianing input.

@zhangpengshan zhangpengshan added this to the Shifu 0.2.7 milestone Jun 1, 2015
@zhangpengshan
Copy link
Contributor Author

Parquet format data is added into Shifu 0.2.7 branch.
When set 'isParquet' to true in ModelConfig#normalize, then run shifu norm and shifu train, data will be saved and read as parquet format.

@zhangpengshan
Copy link
Contributor Author

If norm by parquet format ('isParquet'=true) but train as non parquet('isParquet'=false). Exception will be throw in trainning step, please be aware that should be consistent.

@zhangpengshan
Copy link
Contributor Author

I wouldn't close this ticket now as I need to do several refactor work

@zhangpengshan
Copy link
Contributor Author

Although is is faster to read parquet data in training process(not too much if drop 5% variables), but more time spent on norm step.
Space saving is about 5X.

Done in develop-0.2.7

Close this issue but need further improve speed in norm

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant