Tableau for data mining

I used Churn Modelling from Visualisation > How to use Tableau for Data Mining, a data set of 10000 rows of bank customers data for a simple data mining in Tableau. Analyse bank customer data (6 months period data) and find out why the customers are leaving the bank, using Tableau Public.

note: in this data set the "Exited" column: value 1 means not longer a customer / value 0 means still a customer.

1. First steps:

• Download the .xlsx file in the section above;
• Login into Tableau Public;
• Click on My Profile;
• Click on Create a Viz (a window should appear);
• Click on Upload from computer and choose the csv file you downloaded. The schema of your data should appear in the Tables area (left);

note: The fields marked with blue symbols are the Dimensions and the ones marked with green symbols are the Measures.

2. Create first Tableau sheet (a Map):

• We create an extract of the data set;
• We change the Geography field type to Country/Region by clicking the Geography field then Geography role and choosing Country/Region;
• Add Geography to the tableau working space;
• Add Number of records to the tableau working space;
• Drag Number of records to the Label field;
• Click on the left of the Number of records square (in the working space) and choose Colour (the countries shapes fill with colour);
• Go to the Label square (in the Marks working space) and click on Edit Label. Edit as you please;
• Save (or publish if you are working with Tableau Public);

3. Create second Tableau sheet (making an A/B test * — in this case a gender test — see which gender is more prone of leaving the bank):

• Go to Sheet 2 at the bottom left;
• Move Exited field from Measures to Dimensions;
• Take the Gender field and drag to the Columns field.
• Take the Exited field and drag it to Colour (into the Marks working space). Two different coloured squares should appear in the top right of the working space (one with the label 0 and the second with the label 1);
• Take the Number of records field and drag it into the Rows field. Then add the Number of records field into the Label square in the Marks working space;

We need to transform the Number of records field into a percentage:

• Click the Number of records field in the Marks working space and choose Add Table Calculation;
• Click on Calculation Type and choose Percent of Total;
• Choose Table (down) from Compute Using field;

To format the percentage label:

• Click the Number of records field in the Marks working space choose Format Number , Percentage;
• Click on Decimal Places and choose 0;
• Press control(Windows) or command(Mac), click on Number of records in the Marks working space and drag it to the Rows filed above;

Insights:

From this test we find out that females (25%) are more likely to leave the bank than males (16%).

To validate our insights we need to add a Reference line (adding a Reference line show a bigger picture of the total number of clients that left the bank over a time period, regardless of the gender) :

• Go to Analytics field on the top left;
• From Custom field choose Reference Line and drag it onto the percentage axis (a rectangle with options should appear);
• Drag the Reference Line onto the Table option (other editing rectangle should appear in the working space);
• From the Computation dropdown choose Constant. A number should appear in the Value dropdown menu. Change it to 0.20 (this is the percentage of clients that left the bank regardless of the gender);
• Change the Label field from Value to None;

Extra formating:

• Go to Format (top left) and click on Worksheet (a rectangle with format settings should appear on the right);
• Choose the font and size as you please (this will apply the same format settings to all the worksheet);
• Click on Exited in the Dimensions field and choose Aliases;
• Give 0 and 1 Value (Alias) : Stayed for 0 and Exited for 1; Optional: drag the Exited field from Dimensions to the Label square from the Marks working space.