Alteryx has recently reported the obtaining of Feature Labs, a minor three year-old Cambridge, Mass. Based startup that will include abilities for automating feature engineering to machine learning (ML) models. This was an unadulterated technology acquisition that supplements Alteryx's as of late divulged assisted modeling capacity that guides users through the process of building ML models.
Best known for its self-service data planning capabilities, Alteryx has consistently been a hard company to bind. Is it true that they are just a self-service data prep and visualization tools or more extensive based analytics offering? Contingent upon who you speak to and when, it's either, or both. With the Feature Labs acquisition, Alteryx is multiplying down on analytics and machine learning modeling segment of its portfolio. It's obviously where the goals of their user base – essentially business analysts with yearnings for getting to be citizen data scientists.
In a company blog, Alteryx cited a Kaggle survey indicating feature engineering - where you pick the variables for ML models – was positioned by respondents as the most significant parameter affecting ML model outcomes. The justification is, insofar as you will give a guided experience to building ML models, should get developers and analysts over what they may see as the most critical hump.
Feature Labs, established in 2015, was an outgrowth of MIT explore that up to this point has existed just underneath the radar, having drawn a humble $3 million in funding to date. The small company has three products where the ongoing thread is utilizing ML to enable developers to manufacture ML models.
They incorporate Featuretools, for separating predictive features from data sets. Tempo is a hosted service in the cloud that automates three stages in the development of ML models: prediction engineering, gave a guided experience to issue definition; automated feature engineering, the core family gem; and machine learning, giving an automated workflow to recognizing the best model. The portfolio is balanced with MLApps, which offers a library of prebuilt application templates for issues, for example, predict next buy, predictive maintenance, anti-money laundering, and credit scoring.
The mystery sauce is the thing that the company terms Deep Feature synthesis, a routine imagined at MIT that was adjusted by FeatureLabs cofounders Kalyan Veeramachaneni and Max Kanter to automatically assemble predictive models for complex, multi-table datasets. They guarantee that with this automated function, they beat 66% of the 900 human teams partaking in an online data science rivalry. Given that Alteryx doesn't at present have a hosted SaaS offering, it would intrigue if the Featurelabs Tempo cloud service could turn into the smaller part that manipulates everything else.
Given the humble size of the company, Alteryx didn't uncover the price tag and had the option to finalize the negotiations right away.