(Tee11/Shutterstock)
Access to consistent, high-quality data is one of the toughest challenges in big data, advanced analytics and AI. It's a challenge that JP Morgan's Fusion is addressing with its new Containerized Data service, which provides institutional investors with consistent, enriched data that's standardized with a common semantic layer.
The worst kept secret in big data is that data preparation accounts for most of the time spent in analytics, machine learning, and AI projects. Raw data contains signals that data scientists are desperate to exploit for competitive advantage, but it must be thoroughly cleansed and standardized before it can be combined with other data sets for analysis or fed into machine learning algorithms to train predictive models.
JPMorgan says containerized data will reduce data preparation times and provide higher quality data to meet investors' downstream data needs. The containerized data is delivered through a cloud-based data mesh and data lakehouse service that JPMorgan calls Fusion, with the goal of providing the same look, feel and behavior across sources, regardless of the type of data or analysis being done.
Image courtesy of JP Morgan
“This end-to-end solution leverages a new common semantic layer to model and normalize data across multiple providers, sources, types and structures, delivering consistent, enriched data to investors across business services,” the bank said in a press release last week.
“Fusion ingests, transforms and links data to make it interoperable and ready for AI and ML applications. Investors can access their data in a consistent container at any time using cloud-native channels such as APIs, Jupyter notebooks, Snowflake and Databricks,” the bank said.
Containerized Data handles a variety of JP Morgan and non-JP Morgan data including transactions, benchmarks, holdings, portfolios, public assets, ESG data, ABOR, CBOR, IBOR data, etc. Data ingested into Containerized Data is normalized and reconciled according to data standards set and applied by the various containers within the solution (e.g. custody, middle office, fund accounting, custom containers, etc.).
(Connectus Photo/Shutterstock)
“Fusion's linked data consolidates a complete portfolio overview of each element across clients' portfolios and accounts into a single, panoramic view,” the company said. “All investment data is normalized and linked, making it easy to see custody, middle office and fund accounting data, including public and private assets, in one place.”
Linked data makes it easy for analysts and other end users to explore available data, even if it comes from different domains. When analyzing data, containerized data supports data meshes, allowing individual teams to consume normalized and standardized data in the platform of their choice, including on-premise notebooks like Jupyter or cloud-based platforms like Databricks or Snowflake.
“We understand the nuanced data challenges of institutional investors, and with containerized data we address our clients' most pressing needs,” Jason Mirsky, head of data solutions for securities services at JPMorgan, said in a press release. “Our financial data expertise, vast reference data universe and strategic industry collaborations allow us to model data in ways no one else can and solve our clients' unique data pain points.”
Related Products:
Breaking down silos and building insights: Implementing a data fabric
Data Preparation Market Spotlight: A Look at Tools and Trends
The 7 sins of data preparation
