The state of data scientists: Overwhelmed and underfunded

3 years ago 373

A new survey finds that companies are not making the right investments to support big revenue goals for machine learning initiatives.

Data science initiatives need a strategic makeover to break down silos, support long-term thinking and improve daily operations, according to a new survey.

Three hundred data executives in the U.S. identified a wide range of problems in Domino Data Lab's report, "Data Science Needs to Grow Up: The 2021 Domino Data Lab Maturity Index.

A majority of respondents (82%) were concerned about the impact of both of these issues:

A major revenue loss or a hit to brand reputation stemming from bad or failing models.
A trend toward splashy investments that have short-term payoffs

The survey identified people problems as well, including 44% of survey respondents reporting that they have not hired enough, and about the same amount said they are too siloed off to be effective and have not been given clear roles.

Nick Elprin, CEO and co-founder at Domino Data Lab, said in a press release that executives are not making investments in the right places to support expectations for revenue growth.

"To properly scale data science, companies need to invest in cohesive, sustainable processes to develop, deploy, monitor, and manage models at scale," he said.

SEE: How to become a data scientist: A cheat sheet (TechRepublic)

The survey designed to gauge the state of data science identified these conclusions:

Short-term investment thwarts growth expectations.
The role of data science is unclear.
More revenue requires better models.
Unimproved models bring higher risk.
Teams must clear the obstacles to achieve goals.

The survey also attempted to define profiles for companies with high, increasing and low data maturity models. The survey sample of high maturity companies was small but promising signs included:

Analytics enmeshed in business
Data products drive the organization with robust safeguards
All asset versions are tagged, searchable and reproducible

Challenges with daily operations

The survey found day-to-day challenges as well, starting with getting models into production.

Sixty-eight percent of data executives said that it is somewhat difficult to get models into production to impact business decisions and 37% say it is very to extremely difficult. Maintenance is an issue also with 23% of models never getting an update.

The impact of this failure to follow up goes beyond a wasted investment, according to the survey. A third of data executives said not improving models can result in lost productivity or rework. Also, 43% said not improving models can lead to security or compliance risks, while 41% say it could result in discrimination and bias in modeling.

Finally, 78% of respondents said that they have seen their companies end a data science project or reduce investment if a data model fails, including 26% who said this has happened several times.

According to the survey, the biggest obstacles to success with data-driven work are inadequate data skills among employees; inconsistent standards and processes across the organization; outdated or inadequate tools; lack of buy-in from company leadership; and lack of data infrastructure and architecture.

Wakefield Research conducted the survey for Domino and contacted 300 U.S. executives in data science roles with a minimum seniority of senior director at companies with annual revenue of at least $1 billion. The research was conducted in June 2021 via email invitations and an online survey.