{"id":50373,"date":"2020-03-18T00:00:00","date_gmt":"2020-03-18T00:00:00","guid":{"rendered":"https:\/\/www.techopedia.com\/making-data-analytics-human-for-decision-making\/"},"modified":"2022-03-25T21:23:29","modified_gmt":"2022-03-25T21:23:29","slug":"making-data-analytics-human-for-decision-making","status":"publish","type":"post","link":"https:\/\/www.techopedia.com\/making-data-analytics-human-for-decision-making\/2\/34196","title":{"rendered":"Making Data Analytics Human for Decision-Making"},"content":{"rendered":"
Data science<\/a> needs its Turing test. In executive meetings, decision-makers confer in natural languages and sentiments when they wrestle with the trade-offs of their choices.<\/p>\n They seek to bridge perception differences between them with visuals, charts, infographics and tables.<\/p>\n The amassing of humongous volumes of an expanding variety of variables can potentially transform executive decision-making — when it populates human mind maps.<\/p>\n Better decision-making and data analytics<\/a> are unmistakably a top-of-mind issue for company executives preparing to upgrade their data intelligence<\/a>.<\/p>\n The 2020 State of Data Governance and Automation survey conducted by Dataversity and Erwin reported that better decision-making is the primary driver for their data intelligence initiatives (62% in the 2020 survey up from 45% in 2018) followed by analytics (51%).<\/p>\n Data is agreeable to human decision-makers when it becomes part of a conversation. It engages humans when they gain situational awareness with stories. It stimulates thinking when humans see patterns in visuals. It provokes questions when humans find relationships in the data.<\/p>\n The answers in the data suggest courses of action. Yet, data is formatted to present tables and charts that at best show patterns — unlikely to capture the interdependence in complex environments.<\/p>\n Metadata<\/a>, annotated in human languages, is the bedrock of analytics that communicates narratives. It will start to provide information about events and when, what, where, how, and why they happened.<\/p>\n Humans are spontaneously inclined to choose categories that all their peers can understand without confounding them by varying uses of the terms.<\/p>\n "Historically, the focus of metadata management<\/a> has been on technical metadata (platform, structure, physical characteristics). However, equal weight is now being given to the capture and correlation of business metadata (business rules, associated applications, and business capabilities) and semantic metadata (business terminology and ontology) to support storytelling and contextual association," said Danny Sandwell<\/a>, Director of Product Marketing at erwin.<\/p>\n When data sets<\/a> are as large as they are today, storytellers can put their head around them by looking at a map of the relationships in the data.<\/p>\n Knowledge graphs<\/a>, pioneered by Google<\/a>, shows the interrelationships between entities, their properties, and their relationships. Currently, Google uses knowledge graphs to organize snapshot biographical information about prominent people, such as Donald Trump.<\/p>\n Storytellers are unlikely to be satisfied with the bare-bones data that knowledge graph visualize. For reasons self-evident, they would want to know how Donald Trump came to be the POTUS without any previous background in politics.<\/p>\n They will start to look at his social network and who influenced him, experiences interacting with blue-collar workers and their aspirations, and what he learned from his business failures and successes.<\/p>\n The answers come from databases<\/a> with factual data such as a list of his friends, and other pieces will come from databases with qualitative information such as news.<\/p>\n Community efforts like DBpedia and Yago create a knowledge platform<\/a> with information from Wikipedia to find related information.<\/p>\n Emerging semantic standards like Resource Description Framework<\/a> (RDF) instead use triplets: subject, predicate, and object, which construct snippets like John (subject) lives in the (predicate) suburbs (Object).<\/p>\n Information classified in classes and concepts<\/a>, common across datasets, paves the way for interlinking and finding related information<\/a> and the drivers of variation in either one or both of them. A query language<\/a> like SPARQL<\/a> can search the contents of enormous datasets stored in RDF.<\/p>\n Identifying patterns is only the start of a journey for the exploration of data. A deeper dive reveals stories that executives can use to design their strategies and operational modus operandi to realize the efficiencies to gain competitive advantage.<\/p>\n "Knowledge graphs and an ontology forms the foundation of a data storytelling system, but it needs reasoning and learning layered on top to reach its fullest potential. A storytelling system needs to understand the types of questions users typically ask, the information that is useful to include in the answer, and the types of questions that are likely to be raised by an answer to an earlier question," said Nate Nichols,<\/a> Distinguished Principal, Product Strategy and Architecture of Narrative Science.<\/p>\nNatural Language Data Searches<\/span><\/h2>\n
Finding Data Patterns<\/span><\/h2>\n
Narratives<\/span><\/h2>\n