Skip to main content

A django model to persist and track machine learning models

Project description


Tidy Persistence and Retrieval for Machine Learning


Django-Estimators helps persist and track machine learning models (aka estimators) and datasets.

This library provides a series of proxy objects that wrap common python machine learning objects and dataset objects. As a result, this library can be used to version, track progress and deploy models. It’s highly extensible and can be used with almost any python object (scikit-learn, numpy arrays, modules, methods).

This repo utilizes django as an ORM. If you’d like to work outside of django, try the sqlalchemy-based estimators library instead.


Django-estimators is on PyPI, so just run:

pip install django-estimators

Quick start

1. Add “estimators” to your INSTALLED_APPS django setting like this


2. To create the estimators table, run

python migrate

3. Run python shell and get create new models like so

from sklearn.ensemble import RandomForestClassifier
rfc = RandomForestClassifier()

from estimators.models import Estimator
est = Estimator(estimator=rfc)
est.description = 'a simple forest'

4. Retrieve your model, using the classic django orm, we can pull the last Estimator

est = Estimator.objects.last()
# now use your estimator

Use Case: Retrieving Models/Estimators

If you aren’t sure if it exists, the recommended method is to use the get_or_create method

est = Estimator.objects.get_or_create(estimator=object)
# or potentially update it with update_or_create
est = Estimator.objects.update_or_create(estimator=object)

If you already have the model, in this case of type object

est = Estimator.objects.filter(estimator=object).first()

If you know the unique hash of the model

est = Estimator.objects.filter(object_hash='d9c9f286391652b89978a6961b52b674').first()

Use Case: Persisting and Retrieving DataSets

The DataSet class functions just like the Estimator class. If you have a numpy matrix or a pandas dataframe, you can wrap it with a DataSet object

import numpy as np
import pandas as pd

df = pd.DataFrame(np.random.randint(0,10,(100,8)))

from estimators.models import DataSet

ds = DataSet(data=df)

You can pull that same DataSet object later with

ds = DataSet.objects.latest('create_date')

And if you already have the dataset

ds = DataSet.objects.filter(data=df).first()

Use Case: Persisting Predictions and Results

Sometimes the most valuable part of a machine learning is the whole process. Using an Evaluator object, we can define the relationships between X_test, y_test and y_predicted ahead of time.

Then we can evaluate the evaluation plan, which in turn calls the predict method on the estimator and then presists all the wrapped objects.

Here’s a demo of using an Evaluator.

from sklearn.datasets import load_digits
from sklearn.ensemble import RandomForestClassifier

digits = load_digits() # 1797 by 64
X =
y =

# simple splitting for validation testing
X_train, X_test = X[:1200], X[1200:]
y_train, y_test = y[:1200], y[1200:]

rfc = RandomForestClassifier(), y_train)

Now create your evaluation plan

from estimators.models import Evaluator
plan = Evaluator(X_test=X_test, y_test=y_test, estimator=rfc)

result = plan.evaluate() # executes `predict` method on X_test

And you can view all the atributes on the evaluation result

result.y_test # optional, used with supervised classifiers

Using with Jupyter Notebook (or without a django app)

Django-Estimators can run as a standalone django app.In order to have access to the django db, you’ll need to set up the environment variable to load up your django project. In ipython, by default you can set the environment variable DJANGO_SETTINGS_MODULE to estimators.template_settings like so

import os
import django
os.environ['DJANGO_SETTINGS_MODULE'] = "estimators.template_settings"

If you’re creating a new database (by default it’s db.sqlite3). Therefore we need to run migrations, so in python

from import call_command

Now you can continue you as usual…

from estimators.models import Estimator

To use your own custom settings, make a copy of the estimators.template_settings and edit the fields. Like above, run os.environ['DJANGO_SETTINGS_MODULE'] = "custom_settings_file" before running django.setup().

Development Installation

To install the latest version of django-estimators, clone the repo, change directory to the repo, and pip install it into your current virtual environment.:

$ git clone
$ cd django-estimators
$ <activate your project’s virtual environment>
(virtualenv) $ pip install -e .  # the dot specifies for this current repo

Project details

Download files

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

Built Distribution

django_estimators-0.2.1-py2.py3-none-any.whl (20.1 kB view hashes)

Uploaded py2 py3

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page