Both are **columnar (disk-)storage formats** for use in data analysis systems. 
Both are integrated within [Apache Arrow][1] ([pyarrow][2] package for python) and are
designed to correspond with [Arrow][3] as a columnar in-memory analytics layer.

How do both formats differ?

Should you always prefer feather when working with pandas when possible?

What are the use cases where [feather][4] is more suitable than [parquet][5] and the 
other way round?


----------
Appendix

I found some hints here https://github.com/wesm/feather/issues/188,
but given the young age of this project, it&#39;s possibly a bit out of date.


Not a serious speed test because I&#39;m just dumping and loading a whole
Dataframe but to give you some impression if you never
heard of the formats before:

     # IPython    
    import numpy as np
    import pandas as pd
    import pyarrow as pa
    import pyarrow.feather as feather
    import pyarrow.parquet as pq
    import fastparquet as fp

  
    df = pd.DataFrame({&#39;one&#39;: [-1, np.nan, 2.5],
                       &#39;two&#39;: [&#39;foo&#39;, &#39;bar&#39;, &#39;baz&#39;],
                       &#39;three&#39;: [True, False, True]})
    
    print(&quot;pandas df to disk ####################################################&quot;)
    print(&#39;example_feather:&#39;)
    %timeit feather.write_feather(df, &#39;example_feather&#39;)
    # 2.62 ms &#177; 35.8 &#181;s per loop (mean &#177; std. dev. of 7 runs, 100 loops each)
    print(&#39;example_parquet:&#39;)
    %timeit pq.write_table(pa.Table.from_pandas(df), &#39;example.parquet&#39;)
    # 3.19 ms &#177; 51 &#181;s per loop (mean &#177; std. dev. of 7 runs, 100 loops each)
    print()
    
    print(&quot;for comparison:&quot;)
    print(&#39;example_pickle:&#39;)
    %timeit df.to_pickle(&#39;example_pickle&#39;)
    # 2.75 ms &#177; 18.8 &#181;s per loop (mean &#177; std. dev. of 7 runs, 100 loops each)
    print(&#39;example_fp_parquet:&#39;)
    %timeit fp.write(&#39;example_fp_parquet&#39;, df)
    # 7.06 ms &#177; 205 &#181;s per loop (mean &#177; std. dev. of 7 runs, 1 loop each)
    print(&#39;example_hdf:&#39;)
    %timeit df.to_hdf(&#39;example_hdf&#39;, &#39;key_to_store&#39;, mode=&#39;w&#39;, table=True)
    # 24.6 ms &#177; 4.45 ms per loop (mean &#177; std. dev. of 7 runs, 100 loops each)
    print()
    
    print(&quot;pandas df from disk ##################################################&quot;)
    print(&#39;example_feather:&#39;)
    %timeit feather.read_feather(&#39;example_feather&#39;)
    # 969 &#181;s &#177; 1.8 &#181;s per loop (mean &#177; std. dev. of 7 runs, 1000 loops each)
    print(&#39;example_parquet:&#39;)
    %timeit pq.read_table(&#39;example.parquet&#39;).to_pandas()
    # 1.9 ms &#177; 5.5 &#181;s per loop (mean &#177; std. dev. of 7 runs, 1000 loops each)
    
    print(&quot;for comparison:&quot;)
    print(&#39;example_pickle:&#39;)
    %timeit pd.read_pickle(&#39;example_pickle&#39;)
    # 1.07 ms &#177; 6.21 &#181;s per loop (mean &#177; std. dev. of 7 runs, 1000 loops each)
    print(&#39;example_fp_parquet:&#39;)
    %timeit fp.ParquetFile(&#39;example_fp_parquet&#39;).to_pandas()
    # 4.53 ms &#177; 260 &#181;s per loop (mean &#177; std. dev. of 7 runs, 1 loop each)
    print(&#39;example_hdf:&#39;)
    %timeit pd.read_hdf(&#39;example_hdf&#39;)
    # 10 ms &#177; 43.4 &#181;s per loop (mean &#177; std. dev. of 7 runs, 100 loops each)
    
    # pandas version: 0.22.0
    # fastparquet version: 0.1.3
    # numpy version: 1.13.3
    # pandas version: 0.22.0
    # pyarrow version: 0.8.0
    # sys.version: 3.6.3
    # example Dataframe taken from https://arrow.apache.org/docs/python/parquet.html

  [1]: https://arrow.apache.org/
  [2]: https://pypi.python.org/pypi/pyarrow
  [3]: https://arrow.apache.org/docs/python/index.html
  [4]: https://arrow.apache.org/docs/python/ipc.html#feather-format
  [5]: https://arrow.apache.org/docs/python/parquet.html

All Feather Solutions on Gang of Coders

Total of 1 Feather Solutions

What are the differences between feather and parquet?