.. DO NOT EDIT. .. THIS FILE WAS AUTOMATICALLY GENERATED BY SPHINX-GALLERY. .. TO MAKE CHANGES, EDIT THE SOURCE PYTHON FILE: .. "auto_examples/compressors_comparison.py" .. LINE NUMBERS ARE GIVEN BELOW. .. only:: html .. note:: :class: sphx-glr-download-link-note :ref:`Go to the end ` to download the full example code .. rst-class:: sphx-glr-example-title .. _sphx_glr_auto_examples_compressors_comparison.py: =============================== Improving I/O using compressors =============================== This example compares the compressors available in Joblib. In the example, Zlib, LZMA and LZ4 compression only are used but Joblib also supports BZ2 and GZip compression methods. For each compared compression method, this example dumps and reloads a dataset fetched from an online machine-learning database. This gives 3 information: the size on disk of the compressed data, the time spent to dump and the time spent to reload the data from disk. .. GENERATED FROM PYTHON SOURCE LINES 14-19 .. code-block:: Python import os import os.path import time .. GENERATED FROM PYTHON SOURCE LINES 20-25 Get some data from real-world use cases --------------------------------------- First fetch the benchmark dataset from an online machine-learning database and load it in a pandas dataframe. .. GENERATED FROM PYTHON SOURCE LINES 25-37 .. code-block:: Python import pandas as pd url = "https://github.com/joblib/dataset/raw/main/kddcup.data.gz" names = ("duration, protocol_type, service, flag, src_bytes, " "dst_bytes, land, wrong_fragment, urgent, hot, " "num_failed_logins, logged_in, num_compromised, " "root_shell, su_attempted, num_root, " "num_file_creations, ").split(', ') data = pd.read_csv(url, names=names, nrows=1e6) .. GENERATED FROM PYTHON SOURCE LINES 38-42 Dump and load the dataset without compression --------------------------------------------- This gives reference values for later comparison. .. GENERATED FROM PYTHON SOURCE LINES 42-47 .. code-block:: Python from joblib import dump, load pickle_file = './pickle_data.joblib' .. GENERATED FROM PYTHON SOURCE LINES 48-49 Start by measuring the time spent for dumping the raw data: .. GENERATED FROM PYTHON SOURCE LINES 49-55 .. code-block:: Python start = time.time() with open(pickle_file, 'wb') as f: dump(data, f) raw_dump_duration = time.time() - start print("Raw dump duration: %0.3fs" % raw_dump_duration) .. rst-class:: sphx-glr-script-out .. code-block:: none Raw dump duration: 0.168s .. GENERATED FROM PYTHON SOURCE LINES 56-57 Then measure the size of the raw dumped data on disk: .. GENERATED FROM PYTHON SOURCE LINES 57-60 .. code-block:: Python raw_file_size = os.stat(pickle_file).st_size / 1e6 print("Raw dump file size: %0.3fMB" % raw_file_size) .. rst-class:: sphx-glr-script-out .. code-block:: none Raw dump file size: 167.218MB .. GENERATED FROM PYTHON SOURCE LINES 61-62 Finally measure the time spent for loading the raw data: .. GENERATED FROM PYTHON SOURCE LINES 62-68 .. code-block:: Python start = time.time() with open(pickle_file, 'rb') as f: load(f) raw_load_duration = time.time() - start print("Raw load duration: %0.3fs" % raw_load_duration) .. rst-class:: sphx-glr-script-out .. code-block:: none Raw load duration: 0.075s .. GENERATED FROM PYTHON SOURCE LINES 69-74 Dump and load the dataset using the Zlib compression method ----------------------------------------------------------- The compression level is using the default value, 3, which is, in general, a good compromise between compression and speed. .. GENERATED FROM PYTHON SOURCE LINES 76-77 Start by measuring the time spent for dumping of the zlib data: .. GENERATED FROM PYTHON SOURCE LINES 77-84 .. code-block:: Python start = time.time() with open(pickle_file, 'wb') as f: dump(data, f, compress='zlib') zlib_dump_duration = time.time() - start print("Zlib dump duration: %0.3fs" % zlib_dump_duration) .. rst-class:: sphx-glr-script-out .. code-block:: none Zlib dump duration: 0.695s .. GENERATED FROM PYTHON SOURCE LINES 85-86 Then measure the size of the zlib dump data on disk: .. GENERATED FROM PYTHON SOURCE LINES 86-90 .. code-block:: Python zlib_file_size = os.stat(pickle_file).st_size / 1e6 print("Zlib file size: %0.3fMB" % zlib_file_size) .. rst-class:: sphx-glr-script-out .. code-block:: none Zlib file size: 3.943MB .. GENERATED FROM PYTHON SOURCE LINES 91-92 Finally measure the time spent for loading the compressed dataset: .. GENERATED FROM PYTHON SOURCE LINES 92-99 .. code-block:: Python start = time.time() with open(pickle_file, 'rb') as f: load(f) zlib_load_duration = time.time() - start print("Zlib load duration: %0.3fs" % zlib_load_duration) .. rst-class:: sphx-glr-script-out .. code-block:: none Zlib load duration: 0.270s .. GENERATED FROM PYTHON SOURCE LINES 100-105 .. note:: The compression format is detected automatically by Joblib. The compression format is identified by the standard magic number present at the beginning of the file. Joblib uses this information to determine the compression method used. This is the case for all compression methods supported by Joblib. .. GENERATED FROM PYTHON SOURCE LINES 107-114 Dump and load the dataset using the LZMA compression method ----------------------------------------------------------- LZMA compression method has a very good compression rate but at the cost of being very slow. In this example, a light compression level, e.g. 3, is used to speed up a bit the dump/load cycle. .. GENERATED FROM PYTHON SOURCE LINES 116-117 Start by measuring the time spent for dumping the lzma data: .. GENERATED FROM PYTHON SOURCE LINES 117-124 .. code-block:: Python start = time.time() with open(pickle_file, 'wb') as f: dump(data, f, compress=('lzma', 3)) lzma_dump_duration = time.time() - start print("LZMA dump duration: %0.3fs" % lzma_dump_duration) .. rst-class:: sphx-glr-script-out .. code-block:: none LZMA dump duration: 1.886s .. GENERATED FROM PYTHON SOURCE LINES 125-126 Then measure the size of the lzma dump data on disk: .. GENERATED FROM PYTHON SOURCE LINES 126-130 .. code-block:: Python lzma_file_size = os.stat(pickle_file).st_size / 1e6 print("LZMA file size: %0.3fMB" % lzma_file_size) .. rst-class:: sphx-glr-script-out .. code-block:: none LZMA file size: 2.118MB .. GENERATED FROM PYTHON SOURCE LINES 131-132 Finally measure the time spent for loading the lzma data: .. GENERATED FROM PYTHON SOURCE LINES 132-139 .. code-block:: Python start = time.time() with open(pickle_file, 'rb') as f: load(f) lzma_load_duration = time.time() - start print("LZMA load duration: %0.3fs" % lzma_load_duration) .. rst-class:: sphx-glr-script-out .. code-block:: none LZMA load duration: 0.352s .. GENERATED FROM PYTHON SOURCE LINES 140-146 Dump and load the dataset using the LZ4 compression method ---------------------------------------------------------- LZ4 compression method is known to be one of the fastest available compression method but with a compression rate a bit lower than Zlib. In most of the cases, this method is a good choice. .. GENERATED FROM PYTHON SOURCE LINES 148-151 .. note:: In order to use LZ4 compression with Joblib, the `lz4 `_ package must be installed on the system. .. GENERATED FROM PYTHON SOURCE LINES 153-154 Start by measuring the time spent for dumping the lz4 data: .. GENERATED FROM PYTHON SOURCE LINES 154-161 .. code-block:: Python start = time.time() with open(pickle_file, 'wb') as f: dump(data, f, compress='lz4') lz4_dump_duration = time.time() - start print("LZ4 dump duration: %0.3fs" % lz4_dump_duration) .. rst-class:: sphx-glr-script-out .. code-block:: none LZ4 dump duration: 0.117s .. GENERATED FROM PYTHON SOURCE LINES 162-163 Then measure the size of the lz4 dump data on disk: .. GENERATED FROM PYTHON SOURCE LINES 163-167 .. code-block:: Python lz4_file_size = os.stat(pickle_file).st_size / 1e6 print("LZ4 file size: %0.3fMB" % lz4_file_size) .. rst-class:: sphx-glr-script-out .. code-block:: none LZ4 file size: 6.259MB .. GENERATED FROM PYTHON SOURCE LINES 168-169 Finally measure the time spent for loading the lz4 data: .. GENERATED FROM PYTHON SOURCE LINES 169-176 .. code-block:: Python start = time.time() with open(pickle_file, 'rb') as f: load(f) lz4_load_duration = time.time() - start print("LZ4 load duration: %0.3fs" % lz4_load_duration) .. rst-class:: sphx-glr-script-out .. code-block:: none LZ4 load duration: 0.134s .. GENERATED FROM PYTHON SOURCE LINES 177-179 Comparing the results --------------------- .. GENERATED FROM PYTHON SOURCE LINES 179-201 .. code-block:: Python import numpy as np import matplotlib.pyplot as plt N = 4 load_durations = (raw_load_duration, lz4_load_duration, zlib_load_duration, lzma_load_duration) dump_durations = (raw_dump_duration, lz4_dump_duration, zlib_dump_duration, lzma_dump_duration) file_sizes = (raw_file_size, lz4_file_size, zlib_file_size, lzma_file_size) ind = np.arange(N) width = 0.5 plt.figure(1, figsize=(5, 4)) p1 = plt.bar(ind, dump_durations, width) p2 = plt.bar(ind, load_durations, width, bottom=dump_durations) plt.ylabel('Time in seconds') plt.title('Dump and load durations') plt.xticks(ind, ('Raw', 'LZ4', 'Zlib', 'LZMA')) plt.yticks(np.arange(0, lzma_load_duration + lzma_dump_duration)) plt.legend((p1[0], p2[0]), ('Dump duration', 'Load duration')) .. image-sg:: /auto_examples/images/sphx_glr_compressors_comparison_001.png :alt: Dump and load durations :srcset: /auto_examples/images/sphx_glr_compressors_comparison_001.png :class: sphx-glr-single-img .. rst-class:: sphx-glr-script-out .. code-block:: none .. GENERATED FROM PYTHON SOURCE LINES 202-212 Compared with other compressors, LZ4 is clearly the fastest, especially for dumping compressed data on disk. In this particular case, it can even be faster than the raw dump. Also note that dump and load durations depend on the I/O speed of the underlying storage: for example, with SSD hard drives the LZ4 compression will be slightly slower than raw dump/load, whereas with spinning hard disk drives (HDD) or remote storage (NFS), LZ4 is faster in general. LZMA and Zlib, even if always slower for dumping data, are quite fast when re-loading compressed data from disk. .. GENERATED FROM PYTHON SOURCE LINES 212-218 .. code-block:: Python plt.figure(2, figsize=(5, 4)) plt.bar(ind, file_sizes, width, log=True) plt.ylabel('File size in MB') plt.xticks(ind, ('Raw', 'LZ4', 'Zlib', 'LZMA')) .. image-sg:: /auto_examples/images/sphx_glr_compressors_comparison_002.png :alt: compressors comparison :srcset: /auto_examples/images/sphx_glr_compressors_comparison_002.png :class: sphx-glr-single-img .. rst-class:: sphx-glr-script-out .. code-block:: none ([, , , ], [Text(0, 0, 'Raw'), Text(1, 0, 'LZ4'), Text(2, 0, 'Zlib'), Text(3, 0, 'LZMA')]) .. GENERATED FROM PYTHON SOURCE LINES 219-222 Compressed data obviously takes a lot less space on disk than raw data. LZMA is the best compression method in terms of compression rate. Zlib also has a better compression rate than LZ4. .. GENERATED FROM PYTHON SOURCE LINES 222-225 .. code-block:: Python plt.show() .. GENERATED FROM PYTHON SOURCE LINES 226-228 Clear the pickle file --------------------- .. GENERATED FROM PYTHON SOURCE LINES 228-231 .. code-block:: Python import os os.remove(pickle_file) .. rst-class:: sphx-glr-timing **Total running time of the script:** (0 minutes 7.870 seconds) .. _sphx_glr_download_auto_examples_compressors_comparison.py: .. only:: html .. container:: sphx-glr-footer sphx-glr-footer-example .. container:: sphx-glr-download sphx-glr-download-jupyter :download:`Download Jupyter notebook: compressors_comparison.ipynb ` .. container:: sphx-glr-download sphx-glr-download-python :download:`Download Python source code: compressors_comparison.py ` .. only:: html .. rst-class:: sphx-glr-signature `Gallery generated by Sphinx-Gallery `_