pandas_streaming: streaming API over pandas

pandas_streaming aims at processing big files with pandas, too big to hold in memory, too small to be parallelized with a significant gain. The module replicates a subset of pandas API and implements other functionalities for machine learning.

from pandas_streaming.df import StreamingDataFrame
sdf = StreamingDataFrame.read_csv("filename", sep="\t", encoding="utf-8")

for df in sdf:
    # process this chunk of data
    # df is a dataframe
    print(df)

The module can also stream an existing dataframe.

import pandas
df = pandas.DataFrame([dict(cf=0, cint=0, cstr="0"),
                       dict(cf=1, cint=1, cstr="1"),
                       dict(cf=3, cint=3, cstr="3")])

from pandas_streaming.df import StreamingDataFrame
sdf = StreamingDataFrame.read_df(df)

for df in sdf:
    # process this chunk of data
    # df is a dataframe
    print(df)

It contains other helpers to split datasets into train and test with some weird constraints.

Links:

• GitHub/pandas_streaming

• documentation

• Blog