5 perspectives on modern data analytics


Some things don’t change, even during a pandemic. Consistent with previous years, in CIO’s 2021 State of the CIO survey, a plurality of the 1,062 IT leaders surveyed chose “data/business analytics” as the No.1 tech initiative expected to drive IT investment.

Unfortunately, analytics initiatives seldom do nearly as well when it comes to stakeholder satisfaction.

Last year, CIO contributor Mary K. Pratt offered an excellent analysis of why data analytics initiatives still fail, including poor-quality or siloed data, vague rather than targeted business objectives, and clunky one-size-fits-all feature sets. But a number of fresh approaches and technologies are making these pratfalls less likely.

In this bundle of articles from CIO, Computerworld, CSO, InfoWorld, and Network World, you’ll find advice and examples that can help ensure your own analytics efforts deliver the goods. These initiatives tend to resemble dev projects – even when commercial products are involved – and feature the same well-defined goals and iterative cycles that distinguish successful software development outcomes.

To get the big picture, start with the InfoWorld primer “How to excel with data analytics” by contributor Bob Violino. In this crisply written piece, Violino covers all the bases: establishing analytics centers of excellence; the benefits of self-service solutions (such as Tableau or Power BI); the exciting possibilities for machine learning; and the swing toward cloud analytics solutions. Violino expands on that last point in a second article, this one for CIO: “Analytics in the cloud: Key challenges and how to overcome them.” As he observes, the cloud’s scalability and abundant analytics tools may be irresistible, but migrating masses of company data to the cloud and securing it can be a heart-pounding adventure.

New technology invariably incurs new risks. No advancement has had more momentous impact on analytics than machine learning – from automating data prep to detecting meaningful patterns in data – but it also adds an unforeseen hazard. As CSO Senior Writer Lucian Constantin explains in “How data poisoning attacks corrupt machine learning models,” deliberately skewed data injected by malicious hackers can tilt models toward some nefarious goal. The result could be, say, manipulated product recommendations, or even the ability for hackers to infer confidential underlying data.

Copyright © 2021 IDG Communications, Inc.



Source link