Interactive Data Visualization or Visual Analytics
"A picture is worth a thousand words" or in the case of Data Science, we could say "A picture is worth a thousand statistics". Interactive Data Visualization or Visual Analytics has become one of the top trends in transforming business intelligence (BI) as technologies based on Visual Analytics have moved into widespread use.
Conventional Charts and Dashboards show conclusions but not the thinking behind it. The charts do not empower the user to ask questions and get further. Visual Analytics needs to enable Visual exploration where querying, exploring and visualizing data are a single process along with the ability to shift and link multiple visual perspectives. As a result, the brain's ability to process pictures faster than text is leveraged. For an interesting take on the seven essential elements of a Visual Analytics Application refer to "Selecting a Visual Analytics Application" from Tableau.
Gaps in the Current Feature Selection and Model Creation Process
One of the most important steps in the Data Analytics process is Feature Selection where an analyst helps select inputs or features to help the end user understand and explore the data better. The Analyst or Data Scientist performing the feature selection is usually someone with a statistical background having expertise in one or more of the many Analytics platforms like SAS, SPSS, R, etc. The End User on the other hand is someone more familiar with the business and with a limited understanding of various Feature Selection and Modeling Algorithms.
There is a need for interpretability and transparency in the feature selection and model creation process which will require greater interaction between the Analyst and End User. Visual Analytics will need to play an important role to enable this interaction between different groups.
How can Visual Analytics address this gap?
In a sentence, by creating more Interactive model visualizations where the End User can interact with the model and visualize future scenarios by changing input parameters. For an interesting example, of the power of Interactive Visualizations take a look at this NYT interactive visualization in an article on the Ebola Crisis and compare it with the original CDCC spreadsheets which can be found in the same article.
In a similar fashion, it is possible to design a more interactive Feature Selection process between the Analyst/Data Scientist and the End User where the user is empowered to iterate through different feature subsets using Interactive Visualizations like the one above, rather than having to work with a feature subset chosen for him.