Tableau: Getting Up To Speed
Tableau: Getting Up To Speed (GUTS) is a short course on using Tableau to visualise, explore, and communicate findings on small/medium-sized data sets.
Learners will understand the philosophy of Tableau and gain the capability to create interactive visualisations beyond what is possible with commonly available spreadsheet software like Microsoft Excel.
Specific knowledge and capabilities learners will gain/develop include: (i) data aggregation, (ii) visual encoding of data, (iii) creating interactive visualizations and dashboards (with “app-like” behaviours), (iv) performing exploratory data analysis using Tableau’s descriptive analytics functionality, and (v) performing multi-level aggregations and table calculations.
Also, this course is pegged at Level 3 of the Data Visualization TSC under the Skills Framework for Infocomm Technology.
Underlying Principles of Data Visualisation (… and the GUTS of Tableau)
Alright, perhaps some unfamiliar terms were used like “aggregation” and “table calculations”, so you might wonder if it’s going to be applicable to your work. Let’s get a sense of what they are and what Tableau does.
Consider the following dashboard.
Visually Encoding Information
If we look at the visual in the upper left, we know that we could plot that using two columns of 4 rows of data. So, 8 pieces of information in all, 4 of which are numerical.
We take the information in each row, place the Region along the horizontal axis, and create a vertical bar with height equal to Sales. With that we have converted raw categorical and numerical information into a visual.
But typically, raw data does not come in the above form. Most likely, over a year, an organisation might record thousands or millions of records of what Item was sold, in what Quantity, at what Unit Price, to which Customer, by which Profit Centre. So how did we boil things down to 8 pieces of information to create that visual?
Data Aggregation
Basically, we are going to compute a pivot table. If you don’t remember what that is, let’s be more specific.
In this case, we are going to group all those records by Region, and in each group (which has a list of records), compute the sum (total) of Sales. Each group is a piece of information, and each yields another piece of information for the total Sales (for 8 pieces of information).
Details Glossed Over: Strictly, we do not have the Sales figure in the record as stated above, but that is not a problem. We can express the Sales for each line item by taking Quantity × Unit Price. That is a “calculation” (“transformation”) which involves only information from that record. We also do not have the Region, but we can compute that by finding the Region associated with the relevant Profit Centre (by “joining” the table of those records with a table of information on the Profit Centres).
Now that we have described Data Aggregation (and Calculations and Joins). Let’s briefly touch on some of the things that Tableau facilitates that will be useful.
Table Calculations
Sometimes you need to further transform your aggregates. For example:
This is a plot of the Bollinger Bands of stock prices, a “standard illustration” for Table Calculations. The dots provide the closing price and the line in the middle provides a Moving Average. That is a table calculation which is a computation using data across multiple groups. (Or if you think of each row of an aggregation as a record, a computation across multiple records.) Here is a more specific view.
Other examples of Table Calculations include Window Sum/Minimum/Maximum/Standard Deviation. More examples may be found here.
Multi-Level Aggregation
Sometimes one aggregation alone is not enough. This is where multi-level aggregations come in. Let’s consider this by way of example: Average Total Sales of Sales People by Region.
Suppose we have records of every sale by members of our sales team. In order to compute that, we first need to compute the Total Sales for each Sales Person (group by Sales Person and Region, then sum Sales). That’s our first aggregation. Next, we group by Sales Person and compute the average of Total Sales.
Multi-level aggregations are implemented in Tableau as Level of Detail Expressions. While this is a place where drag and drop is not available, they are not hard once one grasps the concept.
Visual Analytics as a Process: Working Backwards
We have actually covered the GUTS of Tableau. (Hence, the name.) Let’s put things together in a process, starting from what your audience wants to make sense of going back to the data.
Hopefully, all this gives a sense of what the course covers. If you are interested, here is the lesson plan.
“Executive Summary”
Tableau is the best (presently) data to pivot tables to visuals machine on the market. It supports the above process extremely well by enabling you to go from data to visuals using a simple drag and drop process, with little (or no) “programming” (describing calculations as might be done in Excel).
You should consider picking it up.
Lesson Plan
The course is structured to run over 7 sessions of 3 hours each. It is dominantly hands on, but we will touch on some of the intellectual foundations of visual analytics.
Lesson 1: An Overview
- The Elements of Tableau
- Fast Run: From Sheets to Dashboard to Stories
Lessons 2 and 3: Business Intelligence Fundamentals
- The Zen of Tableau
- Getting Data into Tableau
- Data to Visuals: The Charts of Tableau
- A Few More Visuals (Route Visualization; Geocoding; Measure Names/Values)
- Organizing and Supplementing Data Sources
Lesson 4: Analytics & Exploring Data
- Descriptive Statistics via the Analytics Pane
- Exploratory Data Analysis
Lesson 5: Interactive Visual Analysis
- Filters and Parameters
- Counterfactual Analysis
- Dashboards and Stories
- Cross-Sheet Interactions
Lesson 6: Beyond the Basics
- More on Calculated Fields
- Level of Detail (LOD) Calculations
- Table Calculations
- Order of Operations
Lesson 7: Dashboard Development Project
Over Lessons 4 through 6, learners will identify a business problem or organizational issue for which value might be added through simulation. By the start of Lesson 7, learners should have already acquired the required data.
The instructor will guide and support learners through: (i) stating the decisions to be made and basis for selection; (ii) articulating how data visualisation supports decision making; (iii) performing calculations on and aggregation of data; (iv) selecting appropriate visual encodings; (vii) making good use of filters/parameters; (viii) creating value-adding interactions; and (ix) creating data-driven presentations.
Price
The price of the course is set at SGD 1800. But stay tuned for information about available grants.
We are working towards obtaining 70% to 100% grants from IMDA for Singaporean learners. These grants are given on a reimbursement basis. 100% grants are available to NSFs, polytechnic students, and university students; 90% for professionals of age 40 and above; and 70% grants are available to professionals. (See page 3 of this for more information on the CITREP+ Grants.)
