Skip to main content

Monitaur Client Library

Project description

Monitaur Client Library

Tested with the following versions of Python:

  1. 3.8.1
  2. 3.7.6
  3. 3.6.10

Install

$ pip install monitaur

Methods

  1. add_model: Adds metadata about the machine learning model to the system.
  2. get_credentials: Retrieves AWS credentials.
  3. record_training_tabular: Sends trained model and anchors data to S3.
  4. record_training_image: Sends trained model to S3.
  5. record_transaction: Sends transaction details to the server.
  6. read_transactions: Retrieves transactions.

Client Library Examples

from monitaur import Monitaur
from monitaur.utils import hash_file


# create monitaur instance
monitaur = Monitaur(
    client_secret="changme",
    base_url="http://localhost:8008",
)

# train model
dataset = loadtxt("./_example/data.csv", delimiter=",")
seed = 7
test_size = 0.1
model_data = train_model(dataset, seed, test_size)
trained_model = model_data["trained_model"]
training_data = model_data["training_data"]
dump(trained_model, open(f"./_example/data.joblib", "wb"))


# add model to api
model_data = {
    "name": "Diabetes Classifier",
    "model_type": "xgboost",
    "model_class": "tabular",
    "library": "xg_boost",
    "trained_model_hash": hash_file("./_example/data.joblib"),  # trained model (None is allowed)
    "production_file_hash": hash_file("./_example/prediction.py"),  # production file used for running inputs through the trained model (None is allowed)
    "feature_number": 8,
    "owner": "Anthony Habayeb",
    "developer": "Andrew Clark",
    "python_version": "3.8.1",
    "ml_library_version": "1.0.2",
    "influences": True,
}
model_set_id = monitaur.add_model(**model_data)

# get aws credentials
credentials = monitaur.get_credentials(model_set_id)

# record training
record_training_data = {
    "credentials": credentials,
    "model_set_id": model_set_id,
    "trained_model": trained_model,
    "training_data": training_data,
    "feature_names": [
        "Pregnancies",
        "Glucose",
        "BloodPressure",
        "SkinThickness",
        "Insulin",
        "BMI",
        "DiabetesPedigreeF",
        "Age",
    ],
    # "re_train": True
}
monitaur.record_training_tabular(**record_training_data)

# record transaction
prediction = get_prediction([2, 84, 68, 27, 0, 26.7, 0.341, 32])
transaction_data = {
    "credentials": credentials,
    "model_set_id": model_set_id,
    "trained_model_hash": hash_file("./_example/data.joblib"),
    "production_file_hash": hash_file("./_example/prediction.py"),
    "prediction": prediction,
    "image": "cat.jpeg",  # required if 'model_class' is  'image'
    "features": {
        "Pregnancies": 2,
        "Glucose": 84,
        "BloodPressure": 68,
        "SkinThickness": 27,
        "Insulin": 0,
        "BMI": 26.7,
        "DiabetesPedigreeF": 0.341,
        "Age": 32,
    },
}
response = monitaur.record_transaction(**transaction_data)
print(response)

# read transactions by passing model_id and/or model_set_id
# both are optional arguments
transactions = monitaur.read_transactions(model_set_id=model_set_id)
print(transactions)

API Examples

requests:

import requests

API_ENDPOINT = "http://localhost:8000"
CLIENT_SECRET = "eaa74a3d715a36ed6d40af3fb9f5916d8205cf2c"
MODEL_SET_ID = "b7f60d02-06c9-418c-943e-cf74fe61d613"

# get access and refresh tokens
tokens = requests.post(
    f"{API_ENDPOINT}/api/auth/?grant_type=client_credentials",
    data={"client_secret": CLIENT_SECRET}
)
access_token = tokens.json()["access"]
refresh_token = tokens.json()["refresh"]

headers = {"Authorization": f"Token {access_token}"}

# get model metadata
model = requests.get(f"{API_ENDPOINT}/api/models/set/{MODEL_SET_ID}", headers=headers)
print(model.json())
model_id = model.json()["id"]

# get transactions
transactions = requests.get(f"{API_ENDPOINT}/api/transactions/?model={model_id}", headers=headers)
for transaction in transactions.json():
    print(f"\n{transaction}")

cURL:

$ curl -X POST http://localhost:8000/api/auth/\?grant_type=client_credentials \
    -d '{"client_secret": "eaa74a3d715a36ed6d40af3fb9f5916d8205cf2c"}' \
    -H 'Content-Type: application/json'

$ curl -X GET "http://localhost:8000/api/models/set/b7f60d02-06c9-418c-943e-cf74fe61d613/" \
    -H "Authorization: Token 54321"

httpie:

$ http --json POST http://localhost:8000/api/auth/\?grant_type=client_credentials \
    client_secret=eaa74a3d715a36ed6d40af3fb9f5916d8205cf2c

$ http GET http://localhost:8000/api/models/set/b7f60d02-06c9-418c-943e-cf74fe61d613/ Authorization:"Token 54321"

History

TBD

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

monitaur-0.4.0.tar.gz (23.4 kB view hashes)

Uploaded Source

Built Distribution

monitaur-0.4.0-py3-none-any.whl (25.1 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page