Enterprises increase their investment in machine learning

Machine learning development is still in its early stages in many enterprises but investment in the technology is on the increase according to a new report.

The report from Algorithmia shows 22 percent of respondents say their companies have been in production with machine learning for a year. However, 50 percent say they spend between eight and 90 days deploying a single machine learning model.

Part of the problem appears to be scale, which 33 percent of survey respondents cite as the primary pain point in their machine learning life cycle. Other issues include version control and the ability to reproduce models (32 percent), and getting executive buy-in (26 percent).

In order to measure the success of their ML projects, 56 percent of respondents say they are using business metrics, such as ROI, reduced customer churn, and product adoption. The same number say they are using statistical metrics, including accuracy, precision, and speed, to measure success. Both metrics point to the importance of proof-of-concept in machine learning initiatives in order to prove the technology to senior management.

ML skills are in demand too, the study finds that 44 percent of companies employ 10 or more data scientists today, up from 17.5 percent in Algorithmia's 2018 report. At the top end, the latest report shows that five percent of companies have more than 1,000 data scientists, whereas only two percent had as many in 2018.

"The findings of our 2020 study are consistent with what we're hearing from customers," says Diego Oppenheimer, CEO at Algorithmia. "Companies are growing their investments in machine learning, and machine learning operationalization is maturing across all industries, but significant room for growth and improvement remains. The model deployment lifecycle needs to continue to be more efficient and seamless for ML teams. Nevertheless, companies with established ML deployment lifecycles are benefiting from measurable results, including cost reductions, fraud detection, and customer satisfaction. We expect these trends to continue as ML technologies and processes arrive to market and are adopted."

The full report is available from the Algorithmia site.

Image credit: Jirsak/depositphotos.com