Download
JDF is a DataFrames serialization format with the following goals, fast save and load times, compressed storage on disk, enabled disk-based data manipulation (not yet achieved), and support for machine learning workloads, e.g. mini-batch, sampling (not yet achieved). JDF stores a DataFrame in a folder with each column stored as a separate file. There is also a metadata.jls file that stores metadata about the original DataFrame. Collectively, the column files, the metadata file, and the folder is called a JDF "file". JDF.jl is a pure-Julia solution and there are a lot of ways to do nifty things like compression and encapsulating the underlying struture of the arrays that's hard to do in R and Python. E.g. Python's numpy arrays are C objects, but all the vector types used in JDF are Julia data types.
Features
- JDF.jl is the Julia pacakge for all things related to JDF
- Documentation available
- Examples available
- Compressed storage on disk
- Enables disk-based data manipulation (not yet achieved)
- Supports machine learning workloads, e.g. mini-batch, sampling (not yet achieved)
Project Samples
