Metadata-Version: 2.1
Name: xingu
Version: 1.6.1
Summary: Automated ML model training and packaging
Author-email: Avi Alkalay <avi@unix.sh>
License: GNU LESSER GENERAL PUBLIC LICENSE
                               Version 3, 29 June 2007
        
         Copyright (C) 2007 Free Software Foundation, Inc. <https://fsf.org/>
         Everyone is permitted to copy and distribute verbatim copies
         of this license document, but changing it is not allowed.
        
        
          This version of the GNU Lesser General Public License incorporates
        the terms and conditions of version 3 of the GNU General Public
        License, supplemented by the additional permissions listed below.
        
          0. Additional Definitions.
        
          As used herein, "this License" refers to version 3 of the GNU Lesser
        General Public License, and the "GNU GPL" refers to version 3 of the GNU
        General Public License.
        
          "The Library" refers to a covered work governed by this License,
        other than an Application or a Combined Work as defined below.
        
          An "Application" is any work that makes use of an interface provided
        by the Library, but which is not otherwise based on the Library.
        Defining a subclass of a class defined by the Library is deemed a mode
        of using an interface provided by the Library.
        
          A "Combined Work" is a work produced by combining or linking an
        Application with the Library.  The particular version of the Library
        with which the Combined Work was made is also called the "Linked
        Version".
        
          The "Minimal Corresponding Source" for a Combined Work means the
        Corresponding Source for the Combined Work, excluding any source code
        for portions of the Combined Work that, considered in isolation, are
        based on the Application, and not on the Linked Version.
        
          The "Corresponding Application Code" for a Combined Work means the
        object code and/or source code for the Application, including any data
        and utility programs needed for reproducing the Combined Work from the
        Application, but excluding the System Libraries of the Combined Work.
        
          1. Exception to Section 3 of the GNU GPL.
        
          You may convey a covered work under sections 3 and 4 of this License
        without being bound by section 3 of the GNU GPL.
        
          2. Conveying Modified Versions.
        
          If you modify a copy of the Library, and, in your modifications, a
        facility refers to a function or data to be supplied by an Application
        that uses the facility (other than as an argument passed when the
        facility is invoked), then you may convey a copy of the modified
        version:
        
           a) under this License, provided that you make a good faith effort to
           ensure that, in the event an Application does not supply the
           function or data, the facility still operates, and performs
           whatever part of its purpose remains meaningful, or
        
           b) under the GNU GPL, with none of the additional permissions of
           this License applicable to that copy.
        
          3. Object Code Incorporating Material from Library Header Files.
        
          The object code form of an Application may incorporate material from
        a header file that is part of the Library.  You may convey such object
        code under terms of your choice, provided that, if the incorporated
        material is not limited to numerical parameters, data structure
        layouts and accessors, or small macros, inline functions and templates
        (ten or fewer lines in length), you do both of the following:
        
           a) Give prominent notice with each copy of the object code that the
           Library is used in it and that the Library and its use are
           covered by this License.
        
           b) Accompany the object code with a copy of the GNU GPL and this license
           document.
        
          4. Combined Works.
        
          You may convey a Combined Work under terms of your choice that,
        taken together, effectively do not restrict modification of the
        portions of the Library contained in the Combined Work and reverse
        engineering for debugging such modifications, if you also do each of
        the following:
        
           a) Give prominent notice with each copy of the Combined Work that
           the Library is used in it and that the Library and its use are
           covered by this License.
        
           b) Accompany the Combined Work with a copy of the GNU GPL and this license
           document.
        
           c) For a Combined Work that displays copyright notices during
           execution, include the copyright notice for the Library among
           these notices, as well as a reference directing the user to the
           copies of the GNU GPL and this license document.
        
           d) Do one of the following:
        
               0) Convey the Minimal Corresponding Source under the terms of this
               License, and the Corresponding Application Code in a form
               suitable for, and under terms that permit, the user to
               recombine or relink the Application with a modified version of
               the Linked Version to produce a modified Combined Work, in the
               manner specified by section 6 of the GNU GPL for conveying
               Corresponding Source.
        
               1) Use a suitable shared library mechanism for linking with the
               Library.  A suitable mechanism is one that (a) uses at run time
               a copy of the Library already present on the user's computer
               system, and (b) will operate properly with a modified version
               of the Library that is interface-compatible with the Linked
               Version.
        
           e) Provide Installation Information, but only if you would otherwise
           be required to provide such information under section 6 of the
           GNU GPL, and only to the extent that such information is
           necessary to install and execute a modified version of the
           Combined Work produced by recombining or relinking the
           Application with a modified version of the Linked Version. (If
           you use option 4d0, the Installation Information must accompany
           the Minimal Corresponding Source and Corresponding Application
           Code. If you use option 4d1, you must provide the Installation
           Information in the manner specified by section 6 of the GNU GPL
           for conveying Corresponding Source.)
        
          5. Combined Libraries.
        
          You may place library facilities that are a work based on the
        Library side by side in a single library together with other library
        facilities that are not Applications and are not covered by this
        License, and convey such a combined library under terms of your
        choice, if you do both of the following:
        
           a) Accompany the combined library with a copy of the same work based
           on the Library, uncombined with any other library facilities,
           conveyed under the terms of this License.
        
           b) Give prominent notice with the combined library that part of it
           is a work based on the Library, and explaining where to find the
           accompanying uncombined form of the same work.
        
          6. Revised Versions of the GNU Lesser General Public License.
        
          The Free Software Foundation may publish revised and/or new versions
        of the GNU Lesser General Public License from time to time. Such new
        versions will be similar in spirit to the present version, but may
        differ in detail to address new problems or concerns.
        
          Each version is given a distinguishing version number. If the
        Library as you received it specifies that a certain numbered version
        of the GNU Lesser General Public License "or any later version"
        applies to it, you have the option of following the terms and
        conditions either of that published version or of any later version
        published by the Free Software Foundation. If the Library as you
        received it does not specify a version number of the GNU Lesser
        General Public License, you may choose any version of the GNU Lesser
        General Public License ever published by the Free Software Foundation.
        
          If the Library as you received it specifies that a proxy can decide
        whether future versions of the GNU Lesser General Public License shall
        apply, that proxy's public statement of acceptance of any version is
        permanent authorization for you to choose that version for the
        Library.
        
Project-URL: Homepage, https://github.com/avibrazil/xingu
Project-URL: Source, https://github.com/avibrazil/xingu
Project-URL: Issues, https://github.com/avibrazil/xingu/issues/new/choose
Project-URL: Pypi, https://pypi.org/project/xingu
Keywords: mlengineering,xingu
Classifier: Intended Audience :: Developers
Classifier: Programming Language :: Python :: 3
Classifier: Operating System :: OS Independent
Classifier: License :: OSI Approved :: GNU Library or Lesser General Public License (LGPL)
Classifier: Development Status :: 5 - Production/Stable
Classifier: Intended Audience :: Science/Research
Classifier: Topic :: Scientific/Engineering :: Information Analysis
Requires-Python: >=3.8
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: pandas
Requires-Dist: python-decouple
Requires-Dist: SQLAlchemy <2
Requires-Dist: tabulate
Requires-Dist: pygit2
Requires-Dist: randomname
Requires-Dist: pyyaml
Requires-Dist: s3path
Requires-Dist: smart-open
Requires-Dist: scikit-learn
Provides-Extra: docs
Requires-Dist: sklearn ; extra == 'docs'

# Xingu for automated ML model training

Xingu is a framework of a few classes that helps on full industrialization of
your machine learning training and deployment pipelines. Just write your
DataProvider class, mostly in a declarative way, that completely controls
your training and deployment pipeline.

Notebooks are useful in EDA time, but when the modeling is ready to become
a product, use Xingu proposed classes to organize interactions with DB
(queries), data cleanup, feature engineering, hyper-parameters optimization,
training algorithm, general and custom metrics computation, estimation
post-processing.

- Don’t save a pickle at the end of your EDA, let Xingu organize a versioned
  inventory of saved models (PKLs) linked and associated to commit hashes and
  branches of your code.

- Don’t save metrics manually and in an informal way. Metrics are first class
  citizens, so use Xingu to write methods that compute metrics and let it
  store metrics in an organized database that can be queried and compared.

- Don’t make ad-hoc plots to understand your data. Plots are important assets
  to measure the quality of your model, so use Xingu to write methods that
  formaly generate versioned plots.

- Do not worry or even write code that loads pre-req models, use Xingu pre-req
  architecture to load pre-req models for you and package them together.

- Don’t save ad-hoc hypermaters after optimizations. Let Xingu store and manage
  those for you in a way that can be reused in future trains.
  
- Don’t change your code if you want different functionality. Use Xingu
  environment variables or command line parameters to strategize your trains.

- Don’t manually copy PKLs to production environments on S3 or other object
  storage. Use Xingu’s deployment tools to automate the deployment step.
  
- Don’t write database integration code. Just provide your queries and Xingu
  will give you the data. Xingu will also maintain a local cache of your data
  so you won’t hammer your database across multiple retrains. Do the same with
  static data files with parquet, CSV, on local filesystem or object storage.
  
- Xingu can run anyware, from your laptop, with a plain SQLite database, to
  large scale cloud-powered training pipelines with GitOps, Jenkins, Docker
  etc. Xingu’s database is used only to cellect training information, it isn´t
  required later when model is used to predict.

## Install
```shell
pip install https://github.com/avibrazil/xingu
```

or

```shell
pip install xingu
```

## Use to Train a Model
Check your project has the necessary files and folders:
```shell
$ find
dataproviders/
dataproviders/my_dataprovider.py
estimators/
estimators/myrandomestimator.py
models/
data/
plots/
```
Train with DataProviders `id_of_my_dataprovider1` and `id_of_my_dataprovider2`, both defined in `dataproviders/my_dataprovider.py`:
```shell
$ xingu \
    --dps id_of_my_dataprovider1,id_of_my_dataprovider2 \
    --databases athena "awsathena+rest://athena.us..." \
    --query-cache-path data \
    --trained-models-path models \
    --debug
```

## Use the API
See the [proof of concept notebook](https://github.com/avibrazil/xingu/blob/main/notebooks/POC%20Use%20Xingu.ipynb) with vairous usage scenarios:

- POC 1. Train some Models
- POC 2. Use Pre-Trained Models for Batch Predict
- POC 3. Assess Metrics and create Comparative Reports
- POC 4. Check and report how Metrics evolved
- POC 5. Play with Xingu barebones
- POC 6. Play with the `ConfigManager`
- POC 7. Xingu Estimators in the Command Line
- POC 8. Deploy Xingu Data and Estimators between environments (laptop, staging, production etc)

## Procedures defined by Xingu

Xingu classes do all the heavy lifting while you focus on your machine learning
code only.

- Class `Coach` is responsible of coordinating the training process of one or
multiple models. You control parallelism via command line or environment
variables.

- Class `Model` implements a standard pipelines for train, train with hyperparam
optimization, load and save pickles, database access etc. These pipelines are
is fully controlled by your DataProvider or the environment.

- Class `DataProvider` is a base class that is constantly queried by the `Model`
to determine how the `Model` should operate. Your should create a class derived
from `DataProvider` and reimplement whatever you want to change. This will
completely change behaviour of `Model` operation in a way that you´ll get a
completelly different model.

    - It is your `DataProvider` that defines the source of training data as SQL
    queries or URLs of parquets, CSVs, JSONs
    - It is your `DataProvider` that defines how multi-source data should be
    integrated
    - It is your `DataProvider` that defines how data should be split into train
    and test sets
    - Your `DataProvider` defines which `Estimator` class to use
    - Your `DataProvider` defines how the `Estimator` should be initialized and
    optimized
    - Your `DataProvider` defines which metrics should be computed, how to
    compute them and against which dataset
    - Your `DataProvider` defines which plots should be created and against
    which dataset
    - See below when and how each method of your `DataProvider` will be called
    by `xingu.Model`
    
- Class `Estimator` is another base class (that you can reimplement) to contain
estimator-specific affairs. There will be an `Estimator`-derived class for an
XGBoostRegressor, other for a CatBoostClassifier, other for a
SciKit-Learn-specific algorithm, including hyperparam optimization logic and
libraries. A concrete `Estimator` class can and should be reused across multiple
different models.

The hierarchical diagrams below expose complete Xingu pipelines with all their
steps. Steps marked with 💫 are were you put your code. All the rest is Xingu
boilerplate code ready to use.

### `Coach.team_train()`:

Train various Models, all possible in parallel.

1. `Coach.team_train_parallel()` (background, parallelism controled by `PARALLEL_TRAIN_MAX_WORKERS`):
    1. `Coach.team_load()` (for pre-req models not trained in this session)
    2. Per DataProvider requested to be trained:
        1. `Coach.team_train_member()` (background):
            1. `Model.fit()` calls:
                1. 💫`DataProvider.get_dataset_sources_for_train()` return dict of queries and/or URLs
                2. `Model.data_sources_to_data(sources)`
                3. 💫`DataProvider.clean_data_for_train(dict of DataFrames)`
                4. 💫`DataProvider.feature_engineering_for_train(DataFrame)`
                5. 💫`DataProvider.last_pre_process_for_train(DataFrame)`
                6. 💫`DataProvider.data_split_for_train(DataFrame)` return tuple of dataframes
                7. `Model.hyperparam_optimize()` (decide origin of hyperparam)
                    1. 💫`DataProvider.get_estimator_features_list()`
                    2. 💫`DataProvider.get_target()`
                    3. 💫`DataProvider.get_estimator_optimization_search_space()`
                    4. 💫`DataProvider.get_estimator_hyperparameters()`
                    5. 💫`Estimator.hyperparam_optimize()` (SKOpt, GridSearch et all)
                    6. 💫`Estimator.hyperparam_exchange()`
                8. 💫`DataProvider.post_process_after_hyperparam_optimize()`
                9. 💫`Estimator.fit()`
                10. 💫`DataProvider.post_process_after_train()`
    2. `Coach.post_train_parallel()` (background, only if `POST_PROCESS=true`):
        1. Per trained Model (parallelism controled by `PARALLEL_POST_PROCESS_MAX_WORKERS`):
            1. `Model.save()` (PKL save in background)
            2. `Model.trainsets_save()` (save the train datasets, background)
            3. `Model.trainsets_predict()`:
                1. `Model.predict_proba()` or `Model.predict()` (see [below](#predict))
                2. 💫`DataProvider.pre_process_for_trainsets_metrics()`
                3. `Model.compute_and_save_metrics(channel=trainsets)` (see [below](#metrics))
                4. 💫`DataProvider.post_process_after_trainsets_metrics()`
            4. `Coach.single_batch_predict()` (see [below](#batch))



<a id='batch'></a>
### `Coach.team_batch_predict()`:

Load from storage and use various pre-trained Models to estimate data from a pre-defined SQL query.
The batch predict SQL query is defined into the DataProvider and this process will query the database
to get it.

1. `Coach.team_load()` (for all requested DPs and their pre-reqs)
2. Per loaded model:
    1. `Coach.single_batch_predict()` (background)
        1. `Model.batch_predict()`
            1. 💫`DataProvider.get_dataset_sources_for_batch_predict()`
            2. `Model.data_sources_to_data()`
            3. 💫`DataProvider.clean_data_for_batch_predict()`
            4. 💫`DataProvider.feature_engineering_for_batch_predict()`
            5. 💫`DataProvider.last_pre_process_for_batch_predict()`
            6. `Model.predict_proba()` or `Model.predict()` (see [below](#predict))
        2. `Model.compute_and_save_metrics(channel=batch_predict)` (see [below](#metrics))
        3. `Model.save_batch_predict_estimations()`


<a id='predict'></a>
### `Model.predict()` and `Model.predict_proba()`:

1. `Model.generic_predict()`
    1. 💫`DataProvider.pre_process_for_predict()` or `DataProvider.pre_process_for_predict_proba()`
    2. 💫`DataProvider.get_estimator_features_list()`
    3. 💫`Estimator.predict()` or `Estimator.predict_proba()`
    4. 💫`DataProvider.post_process_after_predict()` or `DataProvider.post_process_after_predict_proba()`


<a id='metrics'></a>
### `Model.compute_and_save_metrics()`:

Sub-system to compute various metrics, graphics and transformations over
a facet of the data.

This is executed right after a Model was trained and also during a batch predict.

Predicted data is computed before `Model.compute_and_save_metrics()` is called.
By `Model.trainsets_predict()` and `Model.batch_predict()`.

1. `Model.save_model_metrics()` calls:
    1. `Model.compute_model_metrics()` calls:
        1. `Model.compute_trainsets_model_metrics()` calls:
            1. All `Model.compute_trainsets_model_metrics_{NAME}()`
            2. All 💫`DataProvider.compute_trainsets_model_metrics_{NAME}()`
        2. `Model.compute_batch_model_metrics()` calls:
            1. All `Model.compute_batch_model_metrics_{NAME}()`
            2. All 💫`DataProvider.compute_batch_model_metrics_{NAME}()`
        3. `Model.compute_global_model_metrics()` calls:
            1. All `Model.compute_global_model_metrics_{NAME}()`
            2. All 💫`DataProvider.compute_global_model_metrics_{NAME}()`
    2. `Model.render_model_plots()` calls:
        1. `Model.render_trainsets_model_plots()` calls:
            1. All `Model.render_trainsets_model_plots_{NAME}()`
            3. All 💫`DataProvider.render_trainsets_model_plots_{NAME}()`
        2. `Model.render_batch_model_plots()` calls:
            1. All `Model.render_batch_model_plots_{NAME}()`
            3. All 💫`DataProvider.render_batch_model_plots_{NAME}()`
        3. `Model.render_global_model_plots()` calls:
            1. All `Model.render_global_model_plots_{NAME}()`
            3. All 💫`DataProvider.render_global_model_plots_{NAME}()`
2. `Model.save_estimation_metrics()` calls:
    1. `Model.compute_estimation_metrics()` calls:
        1. All `Model.compute_estimation_metrics_{NAME}()`
        2. All 💫`DataProvider.compute_estimation_metrics_{NAME}()`
