Azure AI (Cortana) vs SAP Predictive Analytics

• Being hosted in Azure solves a massive hosting problem
• The language understanding system has the ability to revolutionize many vertical markets
• Integrating with Cortana Analytics was really simple due to easy to understand documentation

Read full review

• It doesn't require you to have a Ph.D. to build models!
• You can use it to address a very large and wide dataset without worrying about sampling.
• Automation is in the product DNA. You can prepare your data, ingest it into the "Kernel", then get insights about what was found, decide to publish it and schedule scoring tasks or model refresh in the same product.

Read full review

• The language understanding service can be a little overwhelming at first trying to understand the differnces between intents and entities

Read full review

• It's extremely hard to get started. We even have a data scientist who, when we put Predictive Analytics in front of them, they could not intuitively create a data model and start analyzing. Even with deep knowledge of data analysis, the interface isn't intuitive and it's hard to get to the point of having imported a data set and start analyzing/predicting
• Our platform team had confusion installing Predictive Analytics, particularly on the BI Platform. Additional complexity came when it came to using APL Libraries (AFL Wrapper or Stored Procedures)
• We are still working on merging datasets. In theory, we get how to do it but we come across all sorts of issues which don't occur in Spotfire (missing records etc.)

Read full review

• Difficult to ascertain the ROI as we are a software house who have developed a module in our application using Cortana. However for companies that use our software I would say the use of sentiment analysis in our application could free up at least 1 full time resource to be used elsewhere in their organisation.

Read full review

• It will potentially save time in forecast creation by giving an algorithm-derived forecast which can then be adjusted vs. starting from scratch. We can call this, a system enabled forecast
• Thresholds can be created based on historical data to flag actuals as in/out of the 'norm', or to limit the scope of an investigation
• We spent a lot of time so far and haven't got very far with it - it's been quite frustrating to use and a lot of time/money has been invested

Read full review