AI in Predictive Analytics

Dec 7, 2018 · 5 min read
Image source

Analytics has long been one of the principal use cases of technology in the enterprise. At the beginning of the Digital Revolution, it quickly emerged as a natural corollary to the use of spreadsheets and databases as systems of record for core business functions. Once corporations began using computers to keep track of what was happening in their businesses, the natural next step was to use them to ask questions of the data they were tracking and storing to see why things were happening and how they could be improved.

Modern business analytics can be said to have first achieved broad use and popularity with the release of lightweight spreadsheet applications such as VisiCalc, which appeared in 1979. This pioneering application that first put digital analytical tools in the hands of business users was soon supplanted by Lotus 1–2–3 and then Microsoft Excel in the 90s. Over this same period, the technology landscape was also changing in many ways. The client-server revolution was underway in the enterprise, as corporations that had previously only used large centralized mainframe-based systems began to adopt smaller unix-based systems such as the Vax and ultimately adopted the PC. Relational databases such as Oracle also emerged during this period and capitalized on this transition.

The Business Intelligence or BI industry emerged out of this environment in the 1990s, as these changes were well underway, with companies like Business Objects, Cognos and MicroStrategy in the lead. BI in this iteration was predominately based on OLAP cubes, a system that allowed for analysis of structured business data stored in multiple databases. These OLAP cubes, or hypercubes, made it possible for planning, analysis and reporting to be performed based on multiple ‘dimensions’ of data, such as Region, Year, Month, Sales, Version, etc. and to combine data from multiple databases in ways that had not previously been possible.

These early BI implementations were incredibly labor intensive, however, requiring engineering talent to write large numbers of SQL queries and scripts. OLAP cubes needed to be pre-modeled, meaning the analytical model to be employed had to be pre-defined and then data from various sources integrated and fit to match the model using ETL tools or custom scripts. Any time a business user wanted to analyze data, they had to special order the model and the data from the IT department. Deploying these systems frequently required hand coded front ends to be added for business users and the systems needed to be actively maintained.

Eventually, many of these business intelligence 1.0 technologies would be absorbed by tech leaders such as Oracle, IBM and Microsoft and their features integrated into industry leading RDMS’s, spreadsheets and enterprise software suites.

The next major generation of analytics occurred more recently with the advent of easy to use dashboard-based tools that allowed for easy integration of a far wider range of data sources, including structured business data utilizing modern, flexible NoSQL databases and unstructured data such as social media feeds, web pages, XML files and documents. Companies such as Qlik and Tableau emerged as new disruptive leaders in this era. As the previous generation of disruptors emerged during the transition to client server, these new companies achieved prominence during the transition to mobile as a dominant platform. As such, ease of use and drag-and-drop workflows were baked into their products and integral to their success. These new systems importantly allowed for ordinary business users to conduct analysis themselves without requiring input, help or permission from the IT department, thus vastly expanding the impact of analytics across all parts of the enterprise.

More recently, we have seen several strong contenders for the next major trend in business analytics. “Big data” has been a buzzword for several years, with big data analytics companies offering packages that operationalize Apache Hadoop and the R programming language to gain insights from massive amounts of unstructured data gathered in ‘data lakes’. Companies such as Cloudera and Hortonworks are major players here.

The other major area of interest in analytics today is artificial intelligence. AI-powered predictive analytics promises to revolutionize analytics and point the way forward for companies to dramatically improve efficiency and better understand their customers. Major players from the big data platform market have been entering this market via acquisition. Cloudera acquired Fast Forward Labs, Microsoft acquired Lobe and Bonsai and Datazen, SAP acquired, Alteryx acquired Yhat and Workday acquired SkipFlag and And these are just a tiny fraction of what has been a furious pace of M&A industry-wide. These acquisitions frequently enable not only AI analytics capabilities but also the drag-and-drop ease of use that was so integral to the success of the dashboard analytics companies.

The true leaders of the new AI analytics market may still be emerging, however. Major funding rounds are being announced on an almost weekly basis for new startups in the space. “Start-up” may be even be a misnomer in the case of many of these companies, as some of them have raised hundreds of millions of dollars, have been in operation for ten years or more and have numerous Fortune 100 multinational customers.

Here are some of the startups we think it’s worthwhile to keep an eye on in the AI predictive analytics space:

  • Anodot: Machine learning powered automated anomaly detection. Anodot finds outliers in massive amounts of time series data and turns them into valuable business insights
  • Canvass: AI-based predictive analytics on industrial operational data gathered from IoT sensors — specialized for automotive, agriculture, energy and mining industries.
  • analytics platform for metadata generated by devices for telco, utility, network optimization for enterprise, government, network provider and smart city applications
  • QuantCube: real-time predictive analytics on large unstructured data sets with a special focus on financial and economic data and investment intelligence
  • Thoughtspot: self service AI-powered analytics dashboard with natural language interface
  • Showpad: Sales analytics that provides recommendations on content, coaching and next step based on seller behavior an
  • AI-powered marketing analytics for mobile app marketers

By Angus Roven,

Neuromation Investor Relations Analyst


Distributed Synthetic Data Platform for Deep Learning…


Written by


Distributed Synthetic Data Platform for Deep Learning Applications

More From Medium

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade