Eystein from Norway
Favorite quote speaking with him
“Make it simple and fast, my IT team was cheering on Tableau after trying it and QlikView. They really hoped we would pick Tableau and I am very happy we did.”
Eystein recently took over a BI team that wanted to accelerate use of data across every area of the bank and subsidiaries. The bank and subsidiaries are heavy users of SAS Enterprise Guide, SAS Web Report Studio and SAS Enterprise Miner.
They have a variety of data sources and data marts throughout the business. There is also a centralized Enterprise Data Warehouse (EDW) effort that is making headway, but has a long list of items ahead of them. In the meantime, SAS users are rapidly creating new data pulls that Eystein would like to use in Tableau. He realizes that the EDW team will never be able to have all the data that is needed by the business, so SAS is a good platform to prep the data for widespread analysis in Tableau.
They recently evaluated fast analytic tools from more than 20 vendors. Tableau and QlikView made the short list. However, after closely evaluating the ease of creating new analyses and dashboards in both Tableau and QlikView, the IT team strongly encouraged the business to go with Tableau. They have licensed both Tableau Desktop and Tableau Server.
Questions and answers
Stephen, based on your experience with Tableau and SAS, how would you use them together in our bank?
SAS is indeed a powerful system. I previously managed the Enterprise Guide and Add-In for Microsoft Office Development and Testing teams at SAS. SAS Enterprise Guide is the best analytic data preparation tool that I have seen. It can easily access many source systems and allow you to combine the data in most any form for your analysis needs. A wonderful SAS strength is the ability to repeatedly aggregate and merge the data at many different levels of detail, all sourced from a variety of systems.
Tableau also has strong data preparation capabilities, better than most BI tools. However, Tableau is database dependent, just like most BI tools, so it can’t address all of the features possible in SAS for data preparation. Additionally, Tableau currently doesn’t offer multi-pass data scanning and write-back for the data enrichment process. This is where SAS is valuable. However, for typical analyst data prep needs, Tableau is sufficient.
Tableau will definitely shine when you start analyzing your data to answer day-to-day business questions. Tableau’s ability to rapidly ask questions and easily see the answers is without equal. Additionally, users can easily open dashboards created by your team and use them as a “launch pad” for further analysis, easily answering questions not addressed by the work of the central BI team.
Stephen, how would you share SAS Enterprise Miner models with Tableau?
My previous experience with using SAS data mining models alongside Tableau went best when we created a “Customer View”. The idea behind this Customer View was to have a single analysis table that could be enriched with SAS Enterprise Miner model results, at the customer level of detail (one record per customer, lots of data items per customer!)
By creating a customer view that could be enriched with advanced analyst insights, data mining models, lifetime value estimates, customer segmentation results and many other analyses written back to this table, we could easily share these insights with line of business users using Tableau.
For example, suppose that I have built a propensity model for buying a new insurance product. After completing this model, I could add this into the nightly run for updating the customer view as a new column, such as “Odds of purchasing new product”. An analyst in the business, could easily filter by this new data item in Tableau as part of a new email campaign. More important, other teams had also enriched this Customer View with other important metrics, such as Customer Support indicators. Using Tableau Quick Filters, a new user of Tableau could filter to customers with a high propensity to buy and no customer service issues for one message in the campaign, but those high propensity to buy customers with recent customer service issues might be excluded or may receive a personal thank you for being a customer instead of receiving the campaign.
Tableau European Customer Conference 2012, Barcelona, Spain
Tags: Analyst · Articles · Conference · Consumer · Data Mining · Data Warehousing · Finance · Graphs · Insights · Lifetime Value · Principles · SAS · Segmentation · Tableau · Thoughts · ToolsNo Comments
Posted on April 2nd, 2012 by