Data quality hampers FS machine learning
Written by Peter Walker
Refinitiv research has revealed that while the overwhelming majority of financial sector executives have deployed machine learning, most also acknowledge that poor-quality data is impeding their ability to fully leverage the technology.
The global survey among 450 c-suite leaders and heads of data science departments - including 161 from Europe - found that 43 per cent said data quality the biggest barrier to adoption, followed by a lack of data availability (38 per cent) and a lack of available data scientist staff (33 per cent).
The research showed that 90 per cent of financial firms are using machine learning, either in multiple areas as a core part of their business (46 per cent) or in pockets (44 per cent), while the remaining 10 per cent of firms that have not yet deployed machine learning are experimenting with it.
The main applications for using machine learning were in risk use cases (82 per cent of respondents), followed by performance analytics and reporting (74 per cent), with alpha generation in third place (63 per cent). Adoption has been primarily driven by extracting better quality information (60 per cent), increased productivity and speed (48 per cent) and cost reduction (46 per cent).
Meanwhile, 62 per cent of respondents plan to hire more data scientists in the future, as banks and asset managers seek to give themselves a data and technology edge over competitors.
Tim Baker, global head of applied innovation at Refinitiv, commented: “Whether it is an increasingly complex regulatory environment, the need to find new sources of alpha, or winning the fight against financial crime, the industry is turning to data and technology, and data scientists are increasingly important as the alchemists charged with turning big data into insight.”
“We see a future of accelerating innovation fuelled by wider availability of powerful cloud-based artificial intelligence and machine learning tools dramatically lowering entry barriers and thus changing the competitive dynamic across the industry,” he continued. “But no financial institution will be able to use the technology successfully unless the underlying data is machine ready.”