Predictor API client

This package provides a PyPi-installable lightweight client application for the Predictor API RESTFull server application. The package implements PredictorApiClient class enabling fast and easy method-based calls to all endpoints accessible on the API. To make working with the client a piece of cake, it provides full-documented example scripts for each of the supported endpoints. For more information about the Predictor API, please read the official readme and documentation.

The full programming sphinx-generated docs can be seen in the official documentation.

Endpoints:

1. predictor endpoints (/predict and /predict_proba)
   1. /predict - calls .predict on the specified predictor.
   2. /predict_proba - calls .predict_proba on the specified predictor.
2. security endpoints (/signup, /login, and /refresh)
   1. /signup - signs-up a new user.
   2. /login - logs-in an existing user (obtains access and refresh authorization tokens).
   3. /refresh - refreshes an expired access token (obtains refreshed authorization access token).

Contents:

1. Installation
2. Configuration
3. Data
4. Examples
5. License
6. Contributors

Installation

pip install predictor-api-client

Configuration

The package provides the following configuration of the PredictorApiClient object during the instantiation:

1. API deployment specific configuration: it supports the configuration of the host (IP address), port (port number), and other settings related to the deployment and operation of the Predictor API (for more information, see the docs/).
2. API client specific configuration: it supports the configuration of the logging (logging_configuration). In this version, the package provides logging of the successful as well as unsuccessful /predict and /predict_proba endpoint calls (for more information, see the docs/).

Data

The full description of the requirements on input/output data (format, shape, etc.) can be found here.

Examples

In general, every time a client is used, the PredictorApiClient class must be instantiated. Next, all endpoint-specific data must be prepared. And finally, the endpoint-specific methods can be called. The full example scripts for each of the supported endpoints are placed at ./examples (simplified examples are shown bellow).

Client instantiation

from pprint import pprint
from http import HTTPStatus
from predictor_api_client.client import PredictorApiClient

# Prepare the predictor API client settings
#
# --------------------------------------------- #
# Must be same as for the running Predictor API #
# --------------------------------------------- #
#
# 1. host (IP address)
# 2. port (port number)
# 3. request verification
# 4. request timeout in seconds
host = "http://127.0.0.1"
port = 5000
verify = True
timeout = 2

# Instantiate the predictor API client
client = PredictorApiClient(host=host, port=port, verify=verify, timeout=timeout)

User sign-up

# This example assumes the presence of the client instantiation code

# TODO: prepare data for a new user (see the API's requirements on the password)
#
# 1. username
# 2. password (e.g. can be generated with https://passwordsgenerator.net/)
username = "<TODO: FILL-IN>"
password = "<TODO: FILL-IN>"

print("\n-- [01] example --")
print(f"Signing-up a new user with username: {username} and password: {password}\n")

# Sign-up a new user

response, status_code = client.sign_up(username, password)

# Check the output
if status_code == HTTPStatus.OK:
    print("Successfully signed-up a new user")
else:
    print(f"The request was unsuccessful ({status_code}): {response}")

print("Response:")
pprint(response)

User log-in

# This example assumes the presence of the client instantiation code

# TODO: prepare data for an existing user (data from: user sign-up)
#
# 1. username
# 2. password
username = "<TODO: FILL-IN>"
password = "<TODO: FILL-IN>"

print("\n-- [02] example --")
print(f"Logging-in an existing user with username: {username} and password: {password}\n")

# Log-in an existing user
response, status_code = client.log_in(username, password)

# Check the output
if status_code == HTTPStatus.OK:
    print("Successfully logged-in an existing user")
else:
    print(f"The request was unsuccessful ({status_code}): {response}")

print("Response:")
pprint(response)

Expired access token refresh

# This example assumes the presence of the client instantiation code

# TODO: prepare data for request authorization (refresh token from: user log-in)
refresh_token = "<TODO: FILL-IN>"

print("\n-- [03] example --")
print("Refreshing an expired access token\n")

# Refresh an expired access token
response, status_code = client.refresh_access_token(refresh_token)

# Check the output
if status_code == HTTPStatus.OK:
    print("Successfully refreshed an expired access token")
else:
    print(f"The request was unsuccessful ({status_code}): {response}")

print("Response:")
pprint(response)

Prediction

# This example assumes the presence of the client instantiation code

import numpy

# TODO: prepare data for request authorization (access token and refresh token)
access_token = "<TODO: FILL-IN>"
refresh_token = "<TODO: FILL-IN>"

# TODO: prepare model identifier
#
# Example:
# model_identifier = "dummy_predictor"
model_identifier = "<TODO: FILL-IN>"

# TODO: prepare predictor data (feature values/labels)
#
# ---------------------------------------------------- #
# Must meet the data requirements of the Predictor API #
# ---------------------------------------------------- #
#
# Example (10 subjects, each having 100 1-D features):
# feature_values = numpy.random.rand(10, 1, 100)
# feature_labels = None
feature_values = "<TODO: FILL-IN>"
feature_labels = None

print("\n-- [04] example --")
print(f"Calling for prediction(s) on a predictor identified with: {model_identifier}\n")

# Make the prediction(s)
#
# Use one of the following:
# 1. client.predict(...)
# 2. client.predict_proba(...)
response, status_code = client.predict(  # or client.predict_proba(...)
    access_token=access_token,
    refresh_token=refresh_token,
    model_identifier=model_identifier,
    feature_values=feature_values,
    feature_labels=feature_labels)

# Check the output
if status_code == HTTPStatus.OK:
    print("Successfully called .predict(...)/.predict_proba(...)")
else:
    print(f"The request was unsuccessful ({status_code}): {response}")

print("Response:")
pprint(response)

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contributors

This package is developed by the members of Brain Diseases Analysis Laboratory. For more information, please contact the head of the laboratory Jiri Mekyska mekyska@vut.cz or the main developer: Zoltan Galaz galaz@vut.cz.