We're pleased to release Analytic Solver V2018, with a new visual editor for “data science workflows” (also called “pipelines”) that include multiple steps, from data transformations and partitions to training and evaluating machine learning models, and scoring new data. By doing so, we’ve further lowered the complexity and cost of doing predictive and prescriptive analytics, especially for business analysts who aren’t data scientists.
Data Science Workflows Created via “Drag and Drop” or Action Recorder
Using the new Workflow tab in Analytic Solver V2018, you can easily create multi-stage “data science workflows” that may include steps such as data sampling, transformations, feature selection, partitioning, time series forecasting, training machine learning models, and scoring new data. You can either “drag and drop” icons for each step, connected via arrows on a visual canvas, or you can simply turn on a “recording” feature, carry out each step using Analytic Solver’s existing menu options and dialogs, and the workflow diagram will be created automatically. Each step is linked to an Excel worksheet where results can be inspected.
A single button click is sufficient to re-run the entire workflow on existing data. With a new “Map Features” icon, you can connect a new dataset to the front of the workflow, mapping the initially-used data (variables or features) to the new data; then the entire workflow, including training and evaluation of machine learning models, can be executed automatically on the new data.
Capabilities like these are found in "enterprise" data mining software like RapidMiner and SAS Enterprise Miner. But to get the same capacity and speed we offer in Analytic Solver Data Mining, you'd have to pay five times as much in the first year for RapidMiner (ten times as much each following year), and far more than that for SAS Enterprise Miner -- and you wouldn't get any of the comprehensive optimization and simulation features that you get for free with Analytic Solver Data Mining.
Create App - Tableau Menu Option Activated for Tableau 2018.2
Also in this release, we've activated a new option below the Create App icon on the Analytic Solver Ribbon tab: With this option, you can turn your Excel optimization or simulation model into a Tableau Dashboard Extension. In an upcoming release of Tableau due near the end of Q2 2018, and designated Tableau 2018.2, you'll be able to "drag" this Dashboard Extension onto your Tableau dashboard, connect it to data that you've brought into Tableau and placed on that dashboard -- and your Dashboard Extension optimization or simulation model will run, finding solutions based on Tableau data, and displaying them in chart form!
This new feature, which works in a similar way to the existing Create App - Power BI feature, is documented in a new chapter in the Frontline Solvers V2018 User Guide. But a word of caution: You can only use Create App - Tableau with Tableau 2018.2. At present, this means you would need the "developer preview" of Tableau (not the beta version, which is Tableau 2018.1). When Tableau 2018.2 goes into beta testing, if you are a beta tester you'll be able to use this feature. When Tableau 2018.2 is released, Dashbord Extensions will be available for all Tableau users.