Skip to content

What's new? | getML 1.5 release →

Your ML Suite for relational
and time-series data.

getML Relational Learning unlocks a 10x speed-up potential and superior model performance.
A game changer in predictive applications for enterprise applications.

Request a demo API Reference

Powering the world’s best machine learning pipelines.
From next-gen startups to established enterprises.

Machine Learning for enterprise data

Why you should consider getML

getML introduces new ML algorithms that empower data scientists to achieve superior model performance without the burden of manual feature engineering and building complex feature pipelines.

By generalizing gradient boosting to multi-relational decision trees, getML brings supervised learning to raw relational data, enabling end-to-end prediction pipelines. The getML Suite provides an easy to use Python API, adhering to modern standards.

getML is developed by Code17 GmbH from Leipzig, and is used across industries, from finance and manufacturing to healthcare and beyond.

The better solution for predictive analytics on your enterprise data

  • getML is easy to use helping you deliver
    better models faster

  • Delivered 10x speedup in
    customer projects

  • Up to 65% gain in prediction accuracy
    over baseline models

What's under the hood?

Billions of Features with a few lines of code

getML is a high-performance machine learning software for predictive analytics on relational and time series data. At the heart of getML's innovation are five novel feature learning algorithms that automate manual feature engineering using supervised learning. getML enables the creation of end-to-end prediction pipelines capable of learning from terabytes of raw relational data, achieving unparalleled model accuracy within days not months.

Three magic lines of Python code to learn from billions of features.
Press for more details.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
import getml

# Staging of data
data_container = ... # (6)!

# Defintion of data model
star_schema = ... # (7)!

# Define the feature learners
fast_prop = getml.feature_learning.FastProp() # (1)!
relboost = getml.feature_learning.Relboost() # (2)!

# Define the predictor
xgb_predictor = getml.predictors.ScaleGBMRegressor() # (3)!

# Define the pipeline
pipeline = getml.pipeline.Pipeline( # (4)!
    data_model = star_schema.data_model,
    feature_learners = [fast_prop, relboost],
    predictors = xgb_predictor
)

# Train the feature learnings and the predictor
pipeline.fit(data_container.train) # (5)!

pipeline.predict(data_container.test)

  1. FastProp comes with our Community edition. It is fast and generates a substantial number of important features based on simple aggregations.

  2. Relboost is part our Enterprise edition. A generalization of the gradient boosting algorithm, Relboost can learn really complex interdependencies.

  3. ScaleGBMRegressor is our memory-mapped predictor that can handle big datasets that do not fit into memory.

  4. Pipeline bundles together the data model, feature learners and predictors. Just with this line of code, getML takes care of generation and selection of features, and training of the predictor when the pipeline's fit is called next.

  5. Inspired by libraries like scikit-learn, the fit(), score(), and predict() methods of a pipeline make the machine learning process a breeze.

  6. Container holds the actual data of the population and peripheral tables as well as the train-test-validation split. It smoothly ensures the reproducibility of results.

  7. StarSchema captures the relationships between population and peripheral tables and is our go-to data model abstraction. In contrast, unlimited relationship complexity such as the snowflake schema can be modelled with DataModel.

  • Mission of getML

    Making Data Science Fun Again:

    • No complex infrastructure
    • No feature code
    • No long waiting times

  • Getting started is easy

    1. Check out our user guide.
    2. Install getML Community pip install getml.
    3. Clone getml-demo and explore our example notebooks.

    Need help along the way? Here are our support options!

The right getML flavor for your ML application

Both editions share the same Python API.


Community edition

For anyone who worked with Prophet, tsfresh or FeatureTools and is looking for a more memory and run-time efficient solution. getML Community is the leading open source implementation of the propositionalization framework.

Get started


Enterprise edition

This is your choice if shorter development cycles and unprecedented model accuracy provide a competitive edge to your business. getML Enterprise gives you access to the most advanced Relational Learning algorithms.

Learn more

Interested in getML Relational Learning?

Request a meeting to explore the potential of getML Relational Learning for your business application. Or check out one of our code examples before.

Talk to sales Get started