Visualizations and Dashboards

Data science is a field that combines statistics and computer science to turn vast amounts of data into new insights and knowledge. An important aspect of communicating insights is through visual analytics defined as the “science of analytical reasoning facilitated by interactive visual interfaces.”

One way to create visualizations and dashboards is through Tableau. I am in the process of learning Tableau and I find it to be powerful and easy to use. Within just a few minutes visualizations can be created and integrated into a dashboard. In addition, Tableau can also be used with Python and R creating a very effective suite of data science tools.


Data Cleaning and Plots

Once the data is read into Tableau, data cleaning and plot creation is extremely easy and fast. One aspect of Tableau that I really like is that it automatically distinguishes between categorical and continuous variables. The variable type can be changed and cleaned with a click of a mouse button. Plots can be created simply by dragging and dropping these variables onto a column/row menu.

For example, the sales chart below was made simply by dragging and dropping the sales variable from the Measures tab into the Row menu and the sales date from the Dimensions tab into the Columns menu. Coloring the lines by year is done automatically simply by dragging the year variable onto the Color tab in the Marks menu.  Variables can be filtered really nicely, again, this is done by dragging the variable onto the Filters menu.

Screen Shot 2016-01-14 at 1.42.18 PM


Cross Tabulations

Cross tabulations can also be made simply by dragging and dropping. A useful technique is to highlight the table using colors. This cross tab below shows sales highlighted (from green to pink) according to profit:

Screen Shot 2016-01-14 at 3.17.33 PM


Mapping is also quick and easy in Tableau. The below map was created simply by clicking on city and sales and choosing the map icon in the Show Me menu. Geocoding (converting into latitude/longitude coordinates) is done automatically. A nice feature is that categories that are mapped can be changed dynamically using a menu. This map shows total sales colored by profit:

Screen Shot 2016-01-14 at 3.30.34 PM


Once visualizations are created, they can all be integrated in a dashboard. The wonderful thing about dashboards is that the data can be linked. For example, for this sales dashboard, if I want to explore a category (i.e. furniture, technology, etc.) I click on the category of interest and the data for all of the visualizations is updated automatically.

Here is the one for furniture:

Screen Shot 2016-01-14 at 3.44.46 PM


And here is the one for technology:

Screen Shot 2016-01-14 at 3.44.28 PM