{
    "info": {
        "author": "Khaled Sharif",
        "author_email": "kldsrf@gmail.com",
        "bugtrack_url": null,
        "classifiers": [],
        "description": "# sklearn2docker\n#### Convert your trained scikit-learn classifier to a Docker container with a pre-configured API\n\n[![License: LGPL v3](https://img.shields.io/badge/License-LGPL%20v3-blue.svg)](http://www.gnu.org/licenses/lgpl-3.0)\n\n## Installation\n\nThe easiest way to install `sklearn2docker` with all its dependencies is through `pip`:\n\n```bash\npip install git+git://github.com/KhaledSharif/sklearn2docker.git\n```\n\n## Getting started\n\nFirst, create your `sklearn` classifier. In this example we will use the [Iris dataset](http://scikit-learn.org/stable/auto_examples/datasets/plot_iris_dataset.html).\n\n```python\nfrom pandas import DataFrame\nfrom sklearn.datasets import load_iris\nfrom sklearn.tree import DecisionTreeClassifier\n\niris = load_iris()\ninput_df = DataFrame(data=iris['data'], columns=iris['feature_names'])\nclf = DecisionTreeClassifier(max_depth=2)\nclf.fit(input_df.values, iris['target'])\n```\n\nSecond, import the `Sklearn2Docker` class and use it to build your container.\n\n```python\nfrom sklearn2docker.constructor import Sklearn2Docker\n\ns2d = Sklearn2Docker(\n    classifier=clf,\n    feature_names=iris['feature_names'],\n    class_names=iris['target_names'].tolist()\n)\ns2d.save(name=\"classifier\", tag=\"iris\")\n```\n\nThe name and tag arguments we passed to the `save` function are the name and tag of the Docker container we just built ([see: `docker tag`](https://docs.docker.com/engine/reference/commandline/tag/)). Below is an example of the output of the `s2d.save()` line we executed above.\n\n```\nNow attempting to run the command: \n[docker build --file /tmp/tmpywbu3_ad/Dockerfile \n --tag classifier:iris /tmp/tmpywbu3_ad]\n=====================================================================\n> Sending build context to Docker daemon\n> Step 1/6 : FROM python:3.6\n> ---> c1e459c00dc3\n... output truncated ...\n> Step 6/6 : ENTRYPOINT python /code/api.py\n> ---> Running in bd61983358d9\n> Removing intermediate container bd61983358d9\n> ---> fa2041ac6d60\n> Successfully built fa2041ac6d60\n> Successfully tagged classifier:iris\n=====================================================================\nSuccess! You can now run your Docker container using the following command:\n\t docker run -d -p 5000:5000 classifier:iris\n```\n\nYou can now test your container by asking it to predict the same Iris dataset and return the predicted probabilities ([see: `predict_proba`](http://scikit-learn.org/stable/modules/generated/sklearn.tree.DecisionTreeClassifier.html#sklearn.tree.DecisionTreeClassifier.predict_proba)) as a DataFrame.\n\n```python\nfrom os import system\nsystem(\"docker run -d -p 5000:5000 classifier:iris && sleep 5\")\n\nfrom requests import post\nfrom pandas import read_json\nrequest = post(\"http://localhost:5000/predict_proba/split\", json=input_df.to_json(orient=\"split\"))\nresult = read_json(request.content.decode(), orient=\"split\")\nprint(result.head())\n```\n\n```\n   setosa  versicolor  virginica\n0       1         0.0        0.0\n1       1         0.0        0.0\n2       1         0.0        0.0\n3       1         0.0        0.0\n4       1         0.0        0.0\n```\n\nYou can also request regular classification ([see: `predict`](http://scikit-learn.org/stable/modules/generated/sklearn.tree.DecisionTreeClassifier.html#sklearn.tree.DecisionTreeClassifier.predict)). The format for the URL for your Docker container is as so:\n\n```\nhttp://[a]:[b]/[c]/[d]\n\na: the hostname of the container, defaults to `localhost`\nb: the port of the container, defaults to 5000\nc: one of `predict` or `predict_proba`, similar to the scikit-learn api\nd: defaults to `split`; orient of the Pandas DataFrame JSON conversion*\n```\n\n(*: see [this documentation article](https://pandas.pydata.org/pandas-docs/stable/generated/pandas.DataFrame.to_json.html) for more information about Pandas orients, and [this Github issue](https://github.com/pandas-dev/pandas/issues/18912#issuecomment-354430046) for a comparison; most of the time, setting the orient to `split` should do just fine)\n\n```python\nrequest = post(\n    \"http://localhost:5000/predict/split\", \n    json=input_df.to_json(orient=\"split\")\n)\n```\n\n```\n  prediction\n0     setosa\n1     setosa\n2     setosa\n3     setosa\n4     setosa\n```",
        "description_content_type": null,
        "docs_url": null,
        "download_url": "",
        "downloads": {
            "last_day": -1,
            "last_month": -1,
            "last_week": -1
        },
        "home_page": "https://github.com/KhaledSharif/sklearn2docker",
        "keywords": "machine-learning docker deployment data-science",
        "license": "",
        "maintainer": "",
        "maintainer_email": "",
        "name": "sklearn2docker",
        "package_url": "https://pypi.org/project/sklearn2docker/",
        "platform": "",
        "project_url": "https://pypi.org/project/sklearn2docker/",
        "project_urls": {
            "Homepage": "https://github.com/KhaledSharif/sklearn2docker"
        },
        "release_url": "https://pypi.org/project/sklearn2docker/0.1/",
        "requires_dist": null,
        "requires_python": "",
        "summary": "Convert your trained scikit-learn classifier to a Docker container with a pre-configured API.",
        "version": "0.1"
    },
    "last_serial": 3581598,
    "releases": {
        "0.1": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "97375f4724d4d15802881e34c1af8b8a",
                    "sha256": "90781f29d8e8a4accb6568900b272c8655577587cf9f2b6404b4630e8a086c88"
                },
                "downloads": -1,
                "filename": "sklearn2docker-0.1.tar.gz",
                "has_sig": false,
                "md5_digest": "97375f4724d4d15802881e34c1af8b8a",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 3075,
                "upload_time": "2018-02-14T14:52:10",
                "url": "https://files.pythonhosted.org/packages/12/21/01be0cefc165e649bf5d1c0a144f78da550d97837e1a00fe5d05710e773e/sklearn2docker-0.1.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "97375f4724d4d15802881e34c1af8b8a",
                "sha256": "90781f29d8e8a4accb6568900b272c8655577587cf9f2b6404b4630e8a086c88"
            },
            "downloads": -1,
            "filename": "sklearn2docker-0.1.tar.gz",
            "has_sig": false,
            "md5_digest": "97375f4724d4d15802881e34c1af8b8a",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 3075,
            "upload_time": "2018-02-14T14:52:10",
            "url": "https://files.pythonhosted.org/packages/12/21/01be0cefc165e649bf5d1c0a144f78da550d97837e1a00fe5d05710e773e/sklearn2docker-0.1.tar.gz"
        }
    ]
}