Endor emerges from MIT research with unique predictive analytics tech

Endor, a stealth Israeli predictive analytics company, has its roots in some interesting research on human behavior conducted at MIT’s legendary Media Lab.

The company has developed a predictive analytics cloud service based on a concept called ‘Social Physics‘, which purports to simplify big data analysis. The thinking is that people tend to behave in predictable ways, and if you analyze the data through this social prism using formulas based on Social Physics theory, you can generate more accurate results.

Social Physics was a term originally coined by MIT professor Alex “Sandy” Pentland and Endor CTO and co-founder Dr. Yaniv Altshuler (who did his post-doc work at MIT). They found through their research that “…human behavior is determined as much by the patterns of our culture as by rational, individual thinking. These patterns can be described mathematically, and used to make accurate predictions,” Pentland explained in a statement.

Most companies today are generating huge amounts of data, says Endor CEO Doron Alter, but they face challenges building sophisticated models to make accurate predictions based on that data. What’s more, these models tend to be rather rigid, built to handle a single set of questions. If those questions change, it can take weeks or months to adjust.

Endor applied its own proprietary technology based on Social Physics theory, and created a commercial cloud service, where the customer simply feeds the data into the model, and then in minutes (after however it longs to take to upload the data) can begin asking questions and getting answers based on the content of that data in the context of Social Physics. In fact, the biggest bottleneck is not the processing itself. It’s getting the data into the system, according to Alter.

The way it works is that the Endor system looks at the raw data, searching for groups of individuals that display behaviors that Social Physics theory says cannot randomly occur, Alter explained. This is very different from using a generic machine learning algorithm to analyze the same data, he says. “We only look at human behavior using Social Physics equations to transcend the limitations of the traditional machine learning.”

Endor claims that this approach can solve a range of problems for organizations such as a bank finding the most likely candidates who might be looking for a loan or municipalities analyzing taxi data to predict traffic jams.

While there are clearly privacy implications when using data in this way, the company says it’s aware of these ethical issues and their mentor, Professor Pentland, has written extensively on the issues of data privacy and ethical behavior around dealing with big data, which they intend to respect. Still, it’s not hard to see how people could fear that this kind of behavioral analysis could be abused.

Endor raised $5M in seed funding when it launched in September, 2014. That round was led by Marker with participation from Eric Schmidt’s firm, Innovation Endeavors. The company is based in Tel Aviv in Israel and currently has 10 employees. It is still in stealth, working with early customers and design partners on the product, and plans to launch officially at a later date still to be determined.