Deploy AutoML as a service using Flask

Last update: Nov 04, 2022

Related tags

Overview

AutoML Service

Deploy automated machine learning (AutoML) as a service using Flask, for both pipeline training and pipeline serving.

The framework implements a fully automated time series classification pipeline, automating both feature engineering and model selection and optimization using Python libraries, TPOT and tsfresh.

Check out the blog post for more info.

Resources:

TPOT– Automated feature preprocessing and model optimization tool
tsfresh– Automated time series feature engineering and selection
Flask– A web development microframework for Python

Architecture

The application exposes both model training and model predictions with a RESTful API. For model training, input data and labels are sent via POST request, a pipeline is trained, and model predictions are accessible via a prediction route.

Pipelines are stored to a unique key, and thus, live predictions can be made on the same data using different feature construction and modeling pipelines.

An automated pipeline for time-series classification.

The model training logic is exposed as a REST endpoint. Raw, labeled training data is uploaded via a POST request and an optimal model is developed.

Raw training data is uploaded via a POST request and a model prediction is returned.

Using the app

View the Jupyter Notebook for an example.

Deploying

# deploy locally
python automl_service.py

# deploy on cloud foundry
cf push

Usage

Train a pipeline:

train_url = 'http://0.0.0.0:8080/train_pipeline'
train_files = {'raw_data': open('data/data_train.json', 'rb'),
               'labels'  : open('data/label_train.json', 'rb'),
               'params'  : open('parameters/train_parameters_model2.yml', 'rb')}

# post request to train pipeline
r_train = requests.post(train_url, files=train_files)
result_df = json.loads(r_train.json())

returns:

{'featureEngParams': {'default_fc_parameters': "['median', 'minimum', 'standard_deviation', 
                                                 'sum_values', 'variance', 'maximum', 
                                                 'length', 'mean']",
                      'impute_function': 'impute',
                      ...},
 'mean_cv_accuracy': 0.865,
 'mean_cv_roc_auc': 0.932,
 'modelId': 1,
 'modelType': "Pipeline(steps=[('stackingestimator', StackingEstimator(estimator=LinearSVC(...))),
                               ('logisticregression', LogisticRegressionClassifier(solver='liblinear',...))])"
 'trainShape': [1647, 8],
 'trainTime': 1.953}

Serve pipeline predictions:

serve_url = 'http://0.0.0.0:8080/serve_prediction'
test_files = {'raw_data': open('data/data_test.json', 'rb'),
              'params' : open('parameters/test_parameters_model2.yml', 'rb')}

# post request to serve predictions from trained pipeline
r_test  = requests.post(serve_url, files=test_files)
result = pd.read_json(r_test.json()).set_index('id')

example_id	prediction
1	0.853
2	0.991
3	0.060
4	0.995
5	0.003
...	...

View all trained models:

r = requests.get('http://0.0.0.0:8080/models')
pipelines = json.loads(r.json())

{'1':
    {'mean_cv_accuracy': 0.873,
     'modelType': "RandomForestClassifier(...),
     ...},
 '2':
    {'mean_cv_accuracy': 0.895,
     'modelType': "GradientBoostingClassifier(...),
     ...},
 '3':
    {'mean_cv_accuracy': 0.859,
     'modelType': "LogisticRegressionClassifier(...),
     ...},
...}

Running the tests

Supply a user argument for the host.

# use local app
py.test --host http://0.0.0.0:8080

# use cloud-deployed app
py.test --host http://ROUTE-HERE

Scaling the architecture

For production, I would suggest splitting training and serving into seperate applications, and incorporating a fascade API. Also it would be best to use a shared cache such as Redis or Pivotal Cloud Cache to allow other applications and multiple instances of the pipeline to access the trained model. Here is a potential architecture.

A scalable model training and model serving architecture.

Author

Chris Rawles

Deploy AutoML as a service using Flask

Related tags

Overview

AutoML Service

Architecture

Using the app

Deploying

Usage

Running the tests

Scaling the architecture

Author

Owner

Chris Rawles

Backprop makes it simple to use, finetune, and deploy state-of-the-art ML models.

nn-Meter is a novel and efficient system to accurately predict the inference latency of DNN models on diverse edge devices

Automated machine learning: Review of the state-of-the-art and opportunities for healthcare

A collection of machine learning examples and tutorials.

Climin is a Python package for optimization, heavily biased to machine learning scenarios

Laporan Proyek Machine Learning - Azhar Rizki Zulma

fMRIprep Pipeline To Machine Learning

Anomaly Detection and Correlation library

Class-imbalanced / Long-tailed ensemble learning in Python. Modular, flexible, and extensible

Automated Time Series Forecasting

ETNA is an easy-to-use time series forecasting framework.

A repository for collating all the resources such as articles, blogs, papers, and books related to Bayesian Statistics.

Open source time series library for Python

pandas, scikit-learn, xgboost and seaborn integration

Backtesting an algorithmic trading strategy using Machine Learning and Sentiment Analysis.

Machine Learning approach for quantifying detector distortion fields

A mindmap summarising Machine Learning concepts, from Data Analysis to Deep Learning.

Time series changepoint detection

scikit-multimodallearn is a Python package implementing algorithms multimodal data.

Auto updating website that tracks closed & open issues/PRs on scikit-learn/scikit-learn.