How does Design fit into Intelligent Experimentation?

Luis Bermudez and Steven Stein
Experiment Management, Intelligent Experimentation

SigOpt is an Intelligent Experimentation platform designed to accelerate and amplify the impact of modelers everywhere. Modeling is an inherently scientific process, and it is more experimentation-based than traditional software engineering, so it requires extensive experimentation to maximize impact. With this in mind, we need to be able to design experiments that accomplish this goal. That said, do we need to be able to have all the answers from the beginning or do we even need to have all the right questions at the beginning of the experiment? The answer is no. The Intelligent Experimentation platform will help you come up with the right answers and the right questions. “How?” you might ask. Well, it all starts with an initial design of the experiment, and bringing all of the questions, constraints, and metrics to your initial experimental design that you know about and would be relevant to your overarching questions, such as “when to buy or sell?” or “which image do customers enjoy?”. 

Once you start with the initial design, you start exploring the space of the experiment, and learning more about which variables are helping you optimize your goal. At this point, you’ve learned more about your problem space and you might want to re-prioritize a set of parameters or even try a new combination of parameters altogether. You might even have 10 new insights from your first experiment, and thus have 10 new ideas about new experiments you want to run. Well, you’re in luck! Because you can run those 10 new experiments in parallel! And once you complete those 10 new experiments, the Intelligent Experimentation platform can help give you new insights to further scale up your experiment design or prioritize the experiments you are most interested in. The feedback from our dashboard visualizations are going to help you learn more and help you ask new questions so that you can continue to best design your experiments.

Like we mentioned from the start, experimentation is a scientific process and requires many iterations. Similarly, designing your experiments is also an iterative process. Design is not something that you do once and leave it alone. Design is a process where you keep asking better and better questions throughout your experimentation, and SigOpt’s Intelligent Experimentation platform helps you do just that. It helps you ask the right questions through your scientific journey.

To see how SigOpt can improve your design strategy for you and your team, sign up to use it for free

Luis Bermudez AI Developer Advocate
Steven Stein Product Marketing Lead

Want more content from SigOpt? Sign up now.