Get started

Create a free Wyzoo account to have access to all your reports anytime.

Already have an account?

Sign Up

10 min read


A review of Auger.AI Machine Learning platform for use in marketing analytics. is an open source Automated Machine Learning platform designed for creating efficient machine learning models. While machine learning can have a very steep learning curve, Auger aims to remove some of the complexity for those who are not data engineers. As a result it is a tool that could be very useful for direct marketers.

Traditional machine learning typically has required running a tremendous number of painstaking manual tests of algorithms specifically designed for only the subject matter at hand. More advanced tools have focused on creating frameworks which will use the AI itself to train the tool, however these still require a considerable amount of coding, even if only configuring the correct parameters for the algorithm specified.

Auger operates under the assumption that average users of machine learning tools, while somewhat technically savvy, do not have a detailed understanding of the differences between different algorithms or how they should be optimally configured.

Auger allows users to upload a csv file of test data, and train it against a set of built-in standard models. It provides a visual model manager, which enables users to see the performance of a particular model and to be able to compare these against others, in order to determine which ones are the most accurate.



Data Preprocessing

Users can upload a csv file of test data, and Auger will run through some basic preprocessing tasks. It automatically identifies data types, and identifies some basic statistics about each feature (fields in the database), such as range and unique values and a graphical representation of the distribution.

If you are working with a marketing dataset, you will likely will have some records where some pieces of data are missing (such as portions of an address). Auger will be able to quickly help you resolve some of this, and can help weed out highly correlated or low-variance variables.

This will help speed up the performance so that the engine does not need to test against unnecessary attributes and will also remove outliers which could throw off the performance of a model, and provide misleading results.

Auger provides some useful configuration tools for training the models to recognize specific types of data. Here are a few examples:

  • Scaling – for many common algorithms (SVM, k-Nearest Neighbor, support vector machine, logistic regression etc.) you can scale features before training, typically to range between 0 and 1
  • Cyclic Features – you can configure the models to recognize that, for example, the 31st day of a month is one day apart from the 1st day of the following month
  • Categorical Features – for example, synonyms, etc. (M/F, or Male/Female – often set as boolean, but can be configured to recognize “unidentified” characteristics)
  • Sparse Features – for example, a large amount of data that is set as null, but not entirely (e.g. responses to a direct mailing)
  • Feature Interaction – if one type of data within one feature is dependent or affects another.
main types of charts that can be generated by knime


Auger provides a useful leaderboard that allows you to see all of the pipelines that Auger picked up for your dataset, and choose the models which produce the best results and deploy them instantly.

KNIME Analytics has no built-in business intelligence dashboard

Data modeling

Auger performs many standard machine learning tasks, such as classification and regression. It also handles time-series analysis by using regression techniques to overcome the tendency to weight all prior observations equally.

Auger offers some of the state-of-the-art algorithms for model ensembling including:

Model Deployment / API

One of the more useful aspects of Auger is that it has the ability to programmatically run experiments, deploy model pipelines, and generate real-time predictions. Auger has a very simple API, making it relatively easy to use even for those who are not experienced ML engineers.

Wyzoo Star Ratings

3.5 Overall functionality useful to a direct marketer Overall functionality useful to a direct marketer 3.5

Auger can be very useful as an AI engine for marketers, as it has the ability to take a provided test dataset and train it against a large array of existing algorithms. The data gained can be invaluable in using machine learning in helping identify the characteristics of your ideal markets. It is more usable to non-technical types than say, writing code with some existing ML frameworks, however it does appear to require a fairly robust sense of how machine learning works. It's interface is definitely an improvement over not having one at all.

3.5 3.5

For a beginner data scientist: 3.0

for an experienced data scientist: 4.0

While using it is fairly straightforward (assuming you understand what it is you are trying to do), Auger lacks a visualization component This makes it difficult to view data distribtutions to truly understand your results. As visualization is key to understanding your data, this is the main weaknesses of this tool. That said, it does have a fairly robust API, so it should theoretically work well with other applications who can consume and visualize the data processed by Auger.

1.0 1.0

Auger.AI appears to be relatively new. While the Command line interface is public on Github, there does not appear to be much activity apart from a small core group of developers at the company. No forum could be found as of the time of this writing.


  • Commits:


  • Contributors:


  • Releases:


  • Watch:


  • Star:


  • Fork:


  • Commits/Contributors:


3.0 3.0

While one does not need to be a machine-learning engineer to use Auger, it definitely helps to have an understanding of how ML works. The sheer number of different types of algorithms that exist could be difficult to grasp for someone without an innate sense of how AI handles classification and regressions. The fact that it has no internal graphical tools makes it something that cannot stand on its own for a non-technical person. To really use it will require some fairly advanced configuration.

Summary: Key takeaway

Auger.AI is a powerful new tool in the burgeoning field of creating high-level tools for machine learning. It provides a way of training thousands of models against your test dataset and enable you to deploy them directly to your production endpoints.

From the perspective of a direct marketer, Auger can be useful in identifying the nature of who your customers are and to segment them for improving campaign performance. With a rich enough dataset and compared against some large pre-existing models, it can be possible to gain information about your target markets to increase your ability to improve your sales processes.

Auger itself is still somewhat primitive; it's of the first generation of these AI tools for business users. If you are a beginner at ML, data science or data analytics, Auger.Ai can be helpful for your deeper understanding of machine learning concepts. With some work this tool could be very helpful for your direct marketing efforts.


Currently, Auger can only be deployed using AWS or via their own platform.

Want more information about implementing Auger in your organization?

Other News and Reviews