{
    "info": {
        "author": "Igor Gotlibovych",
        "author_email": "igor.gotlibovych@gmail.com",
        "bugtrack_url": null,
        "classifiers": [
            "Intended Audience :: Developers",
            "Operating System :: OS Independent",
            "Programming Language :: Python",
            "Programming Language :: Python :: 3",
            "Programming Language :: Python :: 3.6"
        ],
        "description": "[![PyPI version](https://badge.fury.io/py/deepquantiles.svg)](https://badge.fury.io/py/deepquantiles)\n[![Build Status](https://travis-ci.com/ig248/deepquantiles.svg?branch=master)](https://travis-ci.com/ig248/deepquantiles)\n[![Coverage Status](https://codecov.io/gh/ig248/deepquantiles/branch/master/graph/badge.svg)](https://codecov.io/gh/ig248/deepquantiles)\n\n# Deep Continuous Quantile Regression\nThis package explores different approaches to learning the uncertainty,\nand, more generally, the conditional distribution of the target variable. We introduce a new type of network, the \"Deep Continuous Quantile Regression Network\", that approximates the inverse conditional CDF directly by a mult-layer perceptron, instead of relying on variational methods which require priors on the functional form of the distribution. In many cases we find that it presents a robust alternative to well-known Mixture Density Networks`.\n\n![](https://raw.githubusercontent.com/ig248/deepquantiles/master/README_pics/comparison_good_MDN_good_CDF.png)\n\nThis is particularily important when\n\n- the mean of the target variable is not sufficient for the use case\n- the errors are heteroscedastic, i.e. vary depending on input features\n- the errors are skewed, making a single summary statistic such as variance inadequate.\n\n![](https://raw.githubusercontent.com/ig248/deepquantiles/master/README_pics/comparison_skewed_samples.png)\n\nWe explore two main approches:\n1. fitting a mixture density model\n2. learning the location of conditional qunatiles, `q`, of the distribution.\n\nOur mixture density network exploits an implementation trick to achieve negative-log-likelihood minimisation in `keras`.\n\n![](https://raw.githubusercontent.com/ig248/deepquantiles/master/README_pics/mdn.png)\n\nSame trick is useed to optimize the \"pinball\" loss in quantile regression networks, and in fact can be used to optimize an arbitrary loss function of `(X, y, y_hat)`.\n\nWithin the quantile-based approach, we further explore:\na. fitting a separate model to predict each quantile\nb. fitting a multi-output network to predict multiple quantiles simultaneously\nc. learning a regression on `X` and `q` simultanesously, thus effectively\nlearning the complete (conditional) cumulative density function.\n\n\n## Installation\nInstall package from source:\n\n```\npip install git+https://github.com/ig248/deepquantiles\n```\n\nOr from PyPi:\n\n```\npip install deepquantiles\n```\n## Usage\n```\nfrom deepquantiles import MultiQuantileRegressor, InverseCDFRegressor, MixtureDensityRegressor\n```\nAs this package is largely an experiment, please explore the Jupyter notebooks and expect to look at the source code.\n\n## Content\n- `deepqunatiles.regressors`: implementation of core algorithms\n- `deepquantiles.presets`: a collection of pre-configured estimators and settings used in experiments\n- `deepquantiles.datasets`: functions used for generating test data\n- `deepquantiles.nb_utils`: helper functions used in notebooks\n- `notebooks`: Jupyter notebooks with examples and experiments\n\n## Tests\n\nRun\n```bash\nmake dev-install\nmake lint\nmake test\n```\n\n## References\n**Mixture Density Networks**, Christopher M. Bishop, [NCRG/94/004 (1994)](https://publications.aston.ac.uk/373/1/NCRG_94_004.pdf)\n\n",
        "description_content_type": "text/markdown",
        "docs_url": null,
        "download_url": "",
        "downloads": {
            "last_day": -1,
            "last_month": -1,
            "last_week": -1
        },
        "home_page": "https://github.com/ig248/deepquantiles",
        "keywords": "",
        "license": "MIT",
        "maintainer": "",
        "maintainer_email": "",
        "name": "deepquantiles",
        "package_url": "https://pypi.org/project/deepquantiles/",
        "platform": "",
        "project_url": "https://pypi.org/project/deepquantiles/",
        "project_urls": {
            "Homepage": "https://github.com/ig248/deepquantiles"
        },
        "release_url": "https://pypi.org/project/deepquantiles/0.0.2/",
        "requires_dist": [
            "keras",
            "scikit-learn"
        ],
        "requires_python": "",
        "summary": "Deep Continuous Quantile Regression",
        "version": "0.0.2"
    },
    "last_serial": 4859524,
    "releases": {
        "0.0.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "0d8d6b5b22ca27320363c314268a9118",
                    "sha256": "b38f899fee424729ac2310eda3ee9ba55e59954ed03f0d942e56a5f350b7f170"
                },
                "downloads": -1,
                "filename": "deepquantiles-0.0.0-py2.py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "0d8d6b5b22ca27320363c314268a9118",
                "packagetype": "bdist_wheel",
                "python_version": "py2.py3",
                "requires_python": null,
                "size": 11660,
                "upload_time": "2019-02-16T18:07:09",
                "url": "https://files.pythonhosted.org/packages/4e/cb/8c72f4a82208d04610949bd5394f8db2c31e7c01f69f0b843796bd196675/deepquantiles-0.0.0-py2.py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "6a73547c72cf6f27994da19dbaba248c",
                    "sha256": "9e8d419b3a91d17fe0fe428a1d3ab6e418c27429796e694d855ebdf782121fde"
                },
                "downloads": -1,
                "filename": "deepquantiles-0.0.0.tar.gz",
                "has_sig": false,
                "md5_digest": "6a73547c72cf6f27994da19dbaba248c",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 9164,
                "upload_time": "2019-02-16T18:07:11",
                "url": "https://files.pythonhosted.org/packages/c6/d7/12c742f3e161307412b7f843a319f163d4bbd7f32dfe58d86c201d9b39f7/deepquantiles-0.0.0.tar.gz"
            }
        ],
        "0.0.1": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "76d31348a40f27e865d1b28a768e7fb6",
                    "sha256": "beedd3ea7ecc1564862bc48001cd67c1eff8c4e6c69c9d319cb5bc23d92c325b"
                },
                "downloads": -1,
                "filename": "deepquantiles-0.0.1-py2.py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "76d31348a40f27e865d1b28a768e7fb6",
                "packagetype": "bdist_wheel",
                "python_version": "py2.py3",
                "requires_python": null,
                "size": 11660,
                "upload_time": "2019-02-16T18:09:23",
                "url": "https://files.pythonhosted.org/packages/ce/47/9b573a595e5259b97e453153344e1bd39470ab2a500343df8b2cf5ffd039/deepquantiles-0.0.1-py2.py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "fdc38230b957658030ecbfdc6e596511",
                    "sha256": "ac0e80171714e93d30216f3800b419b651282c781f4a4fb7192baa50c5c433d5"
                },
                "downloads": -1,
                "filename": "deepquantiles-0.0.1.tar.gz",
                "has_sig": false,
                "md5_digest": "fdc38230b957658030ecbfdc6e596511",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 9160,
                "upload_time": "2019-02-16T18:09:25",
                "url": "https://files.pythonhosted.org/packages/c9/7d/7793aa04260fca74e5557db053b46367d8ff5334d84879345a65a77d4be9/deepquantiles-0.0.1.tar.gz"
            }
        ],
        "0.0.2": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "109ad89698838a67d1de51c37ad846fe",
                    "sha256": "a80aa4a121cf23477f8d235a8fcfe0ddb6ae664b21baed5b37315b5f9413aed7"
                },
                "downloads": -1,
                "filename": "deepquantiles-0.0.2-py2.py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "109ad89698838a67d1de51c37ad846fe",
                "packagetype": "bdist_wheel",
                "python_version": "py2.py3",
                "requires_python": null,
                "size": 12042,
                "upload_time": "2019-02-23T23:43:22",
                "url": "https://files.pythonhosted.org/packages/ec/44/954fb3dcdde500d3b7ab7cae302c11d36179ff9d116a88d5c8178e8f38ad/deepquantiles-0.0.2-py2.py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "49a0bd35b040d95e485ff8ac1267945d",
                    "sha256": "91023eec0c9745f34b0fa558b4090aceac4fbe3454836f7c3d45e55ae0dbb0fa"
                },
                "downloads": -1,
                "filename": "deepquantiles-0.0.2.tar.gz",
                "has_sig": false,
                "md5_digest": "49a0bd35b040d95e485ff8ac1267945d",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 9898,
                "upload_time": "2019-02-23T23:43:23",
                "url": "https://files.pythonhosted.org/packages/e1/73/49fb5e90c7176393690020fb2a3d9b4b852ea451ff850fa49855c00ae8e0/deepquantiles-0.0.2.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "109ad89698838a67d1de51c37ad846fe",
                "sha256": "a80aa4a121cf23477f8d235a8fcfe0ddb6ae664b21baed5b37315b5f9413aed7"
            },
            "downloads": -1,
            "filename": "deepquantiles-0.0.2-py2.py3-none-any.whl",
            "has_sig": false,
            "md5_digest": "109ad89698838a67d1de51c37ad846fe",
            "packagetype": "bdist_wheel",
            "python_version": "py2.py3",
            "requires_python": null,
            "size": 12042,
            "upload_time": "2019-02-23T23:43:22",
            "url": "https://files.pythonhosted.org/packages/ec/44/954fb3dcdde500d3b7ab7cae302c11d36179ff9d116a88d5c8178e8f38ad/deepquantiles-0.0.2-py2.py3-none-any.whl"
        },
        {
            "comment_text": "",
            "digests": {
                "md5": "49a0bd35b040d95e485ff8ac1267945d",
                "sha256": "91023eec0c9745f34b0fa558b4090aceac4fbe3454836f7c3d45e55ae0dbb0fa"
            },
            "downloads": -1,
            "filename": "deepquantiles-0.0.2.tar.gz",
            "has_sig": false,
            "md5_digest": "49a0bd35b040d95e485ff8ac1267945d",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 9898,
            "upload_time": "2019-02-23T23:43:23",
            "url": "https://files.pythonhosted.org/packages/e1/73/49fb5e90c7176393690020fb2a3d9b4b852ea451ff850fa49855c00ae8e0/deepquantiles-0.0.2.tar.gz"
        }
    ]
}