Creating a seamless experience for our end users is no longer an option – it’s a requirement. Visualization tools have moved businesses from spreadsheets to charts and now to full on applications. We should be able to provide the same level of service with analytics.
Using a web portal, we can provide reports and collect user inputs in the same space. Using different technologies in harmony, end users can come to same clean interface to run analyses and see the results.
Our portal is hosted in Python-Flask, but the technology is replaceable – a theme throughout this demo. If you would rather use another web framework – you can!
Let’s face it – data scientists would rather embed machine learning and modeling into applications that are self-service than have to constantly rerun code for end users. By using advanced analytics inside of an ecosystem like this, we can free up time for data scientists and put more intelligent data in front of those who need it the most.
The demos here use scripts written in R, Python, and Alteryx modules on Alteryx server to dynamically run analyses. This layer is flexible to the use case – where predictive models need to be re-run from the source data, the Predictive Layer reads data from a data source and writes output in a consumable format for the visual layer (see our Forecasting or HR Demos). In other cases, a model simply needs to read inputs to generate a prediction and can bypass some of those steps. As well, multiple tools can accomplish the same results in the Predictive Layer, making this environment flexible to your organization.
Not only can tools like R, Python, and Alteryx create learning models, but they can reshape the data in a way that makes it easier to consume in tools like Tableau. This layer not only adds predictive analytics capabilities, but allows for dynamic ETL as well.
Storing your organization's data inside a web app probably isn’t manageable. The Data Layer is simply any number of data sources from which this environment can read from and write to. For convenience, it may be easier to read data from one source and store it in another. Either way, the Data Layer allows the application to read in the data it needs, and then store user data in a format that is easy to consume for the Distribution Layer. There is no required data source either. The environment is flexible to what technologies your organization is already using.
In order to provide visualizations and reports to our end users after our Predictive Layer has run, we can utilize tools like Tableau Server, Shiny, and Bokeh to create and host reports to be viewed in the web portal.
When a more robust distribution is need, we use Tableau Server to allow end users to download workbooks and data sets to work with on their own. Tableau also enables faster creation of the visual layer, while also allowing us to use other visualizations alongside like D3.
Knowing before it happens.
Reach the right people at the right time.
Discover and Influence Your Audience
Nuggets aren't always small.
Analyzing the 'Where?'
Letting computers train to win.