English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

The increasing reliance on applications with machine learning (ML) componentscalls for mature engineering techniques that ensure these are built in a robustand future-proof manner. We aim to empirically determine the state of the artin how teams develop, deploy and maintain software with ML components. We minedboth academic and grey literature and identified 29 engineering best practicesfor ML applications. We conducted a survey among 313 practitioners to determinethe degree of adoption for these practices and to validate their perceivedeffects. Using the survey responses, we quantified practice adoption,differentiated along demographic characteristics, such as geography or teamsize. We also tested correlations and investigated linear and non-linearrelationships between practices and their perceived effect using variousstatistical models. Our findings indicate, for example, that larger teams tendto adopt more practices, and that traditional software engineering practicestend to have lower adoption than ML specific practices. Also, the statisticalmodels can accurately predict perceived effects such as agility, softwarequality and traceability, from the degree of adoption for specific sets ofpractices. Combining practice adoption rates with practice importance, asrevealed by statistical models, we identify practices that are important buthave low adoption, as well as practices that are widely adopted but are lessimportant for the effects we studied. Overall, our survey and the analysis ofresponses received provide a quantitative basis for assessment and step-wiseimprovement of practice adoption by ML teams.

Adoption and Effects of Software Engineering Best Practices in Machine Learning