Any questions? Feel completely free to contact us
The goal of analysis isn't only about discovering expected answers; after all, fitting a model to a looked-for solution doesn’t reveal new insights and, ultimately, magnifies the risk of unidentified problems. True predictive analytics is based entirely on the concept of “Discovery”: finding the unexpected, correlating interactions between entities, and discovering patterns that can only emerge from a global perspective.
Graph analytics is the science of visualizing graphs that convey entity interactions, building out metrics based on those graphs, and using those metrics to create more relevant forecasts. Graph analytics is a process that is designed to discover unexpected relationships using a wide variety of datastores, such as social network content, banking details, and publicly-available data from government sources. A proficiency in graph analytics processes means that you’ll be able to analyze interactions, discover patterns, and then apply that information to forecast and solve existing (or potential) problems.
Data Science Studio (DSS) empowers companies to harness graph analytics processes to create their own unique metrics in support of forecasting. This effort starts with data collection: DSS connects to a wide variety of datastores (over 25 connectors) and enables users to take control of the entire data cleansing process (quickly!), from parsing to formatting. DSS sifts through all of the raw data, automatically notifies you of probable errors, and suggests corrections. This data can originate from any source, such as social media channels, publications, websites, electronic health records, financial statements, and even Marvel Superhero data! DSS supports multiple data science packages, including Python Networkx ― a particularly well-suited package for graph analytics (check out our blog to read about one of our data scientists making use of Networkx for this Facebook Recruiting Kaggle Challenge).
Metrics can then be built using a wide variety of network indicators, such as centrality, convergence, and popularity. These indicators are used to help identify useful metrics from graphical analysis, such as point-of-sale terminals used for fraudulent activities across a geographic area, key influencers in a social network, or causes of revenue fluctuations across a business enterprise. Data Science Studio uses these metrics to bring life to the modeling and visualization process, effectively enabling you to connect disparate data points and understand data relationships that were not initially apparent. The DSS platform uses graph analytics to discover patterns of data that can then be further explored to determine veracity and usability for forecasting purposes.
DSS can use graph analytics for application in a wide variety of sectors, such as fraud detection, lifetime value forecast for applications, and trend detection from social networks. Data Science Studio can also use graph analytics methods to develop Web applications or leverage existing graph technologies, such as the D3.js library (samples of D3 visualizations).