[简单使用]

· 读取csv
    import dask.dataframe as dd
    df = dd.read_csv('2015-*-*.csv')
    df.groupby(df.user_id).value.mean().compute()
· 读取hdf5
    import dask.array as da
    f = h5py.File('myfile.hdf5')
    x = da.from_array(f['/big-data'], chunks=(1000, 1000))
    x - x.mean(axis=1).compute()
· Dask Bag
    import dask.bag as db
    b = db.read_text('2015-*-*.json.gz').map(json.loads)
    b.pluck('name').frequencies().topk(10, lambda pair: pair[1]).compute()
· Dask delayed
    from dask import delayed
    L = []
    for fn in filenames:                  # Use for loops to build up computation
        data = delayed(load)(fn)          # Delay execution of function
        L.append(delayed(process)(data))  # Build connections between variables
    result = delayed(summarize)(L)
    result.compute()
· Client
    client = Client(processes=True)
    from_sequence(seq[, partition_size, npartitions]): Create a dask Bag from Python sequence.
    dask.compute(): Compute several dask collections at once.

很好的介绍！