In the Qlik community (we build on their fantastic Qlik Sense platform), there is a lot of anticipation for how Qlik will roll out the AutoML capabilities they acquired with their recent purchase of BigSquid.
Here at Polygon Research, we don’t have to wait – we are actively leveraging Qlik AutoML, as the branding now goes, for explaining historical outcomes as well as predicting future values. So far – we love it. We started using BigSquid’s platform during last year’s version of a data science class we’ve taught since 2020, where they effortlessly scaled to accommodate our 200+ students.
But this raises an upstream question: why AutoML?
AutoML is introduced variously as:
the process of automating the time-consuming, iterative tasks of machine learning model development.
(source)
With AutoML…you don’t need to worry about data preparation, feature engineering, algorithm selection, training and tuning, inference, and continuous model monitoring.
(source)
Train high-quality custom machine learning models with minimal effort and machine learning expertise.
(source)
These snippets focus on time- and talent-related benefits of AutoML (invoking Bahador Khaleghi’s helpful Time-Talent-Trust paradigm), but the primary reason we’re investing in AutoML is the third leg of the stool: trust. In our mission to eliminate information asymmetry in the mortgage industry, we model the industry’s most important open data sets. As we apply ML to better understand the past and predict the future, we want others to be able to both understand and even repeat our results, and we feel that by focusing on 1) asking the right questions and 2) finding the right data – rather than on the myriad steps and choices entailed in building our own machine learning models from scratch – we're improving transparency and, we hope, trust. You’ll hear more about our efforts in this blog series, and we hope to hear your thoughts and responses as well.