My (still short) experience with the DataRobot
Updated: Dec 30, 2019
I’m a data scientist, passionate about data, and recently had the opportunity to try a Machine Learning Automation (AutoML) platform, DataRobot. This platform automates many of the tasks needed to develop artificial intelligence (AI) and machine learning applications and enables more users across an enterprise to succeed with machine learning by simply utilizing their understanding of their data and business and letting DataRobot do the rest.
I admit that, like many professionals in my area, the initial scepticism about the DataRobot platform - and many AutoML platforms - was big, but quickly this feeling changed.
I decided to start a machine learning project on the DataRobot platform with a database I was already familiar with to more easily assess the benefits of the solution. After the data was uploaded, the first interactions with the platform were fantastic. In less than 15 minutes, the platform was able to build and evaluate the performance of more than 80 models using different methodological approaches and pipelines that had been winning formulas for many data science Kaggle competitions. The platform also gives the data scientist the ability to extract documentation for the entire workflow for each model, view pre-uploaded data treatments and analyzes, and extract automatic insights after building the predictive model.
The DataRobot platform offers it all that without the need for code implementation and entirely automatically - I just had to drag and drop, select the target and click START in Run Autopilot mode. As I explored the platform and viewed the early results, I couldn't stop thinking about the time I would need to create and validate over 80 pipelines using just my favourite languages, Python and R. With the help of DataRobot, I also thought how I could start developing data science projects more quickly and safely (error-free), focusing on addressing and framing analytical and business problems, as well as researching and collecting better information for embed in dataset.
After 2 weeks of interacting with the DataRobot platform, I see that my initial scepticism had no reason. My (still short) experience with one of the best AutoML platforms on the market now leads me to say to all data science professionals to embrace with enthusiasm what seems, to me, the inevitable future of our profession: Automation.
See you soon!
Data Scientist @Passio Consulting