I continue my journey with the wonderful tool named Tableau.
Wine and the French have an inseparable relation. Did you in the picturesque region of Bordeaux, France, wine is more than just a beverage; it’s a legacy. The story of Bordeaux’s rise as a wine capital dates back to the marriage of Eleanor of Aquitaine and Henry Plantagenet in 1152, which opened the English market to Bordeaux wines. This union ignited an insatiable English appetite for the region’s distinct reds. The subsequent boom in wine trade transformed Bordeaux, turning its port into a bustling hub of commerce. Centuries later, the wines of Bordeaux remain emblematic of the region’s resilience and innovation, standing as a testament to how a single product can shape the destiny of an entire region.
https://cdn.britannica.com/53/102253-050-F00DEBB6/Eleanor-of-Aquitaine-Louis-VII-Second-Crusade.jpg
However, in our story, the protagonist, Mr. Wine (or Miss. Hey, I am not a misogynist), has travelled many seas to reach the island nation of Australia. Here company has been doing successfully in trading our protagonist. Let us venture into assessing the business from its data.
Our storyboard input consists of nearly 21 features or columns and nearly 10000 data points or rows. Every story can be uniquely told if a narrator reads between the lines and add a flavour of his/her imagination. Let us create one such derivative, a new column (feature engineering) from two columns which is the total Revenue (unit price * order quantity).
Formatting a value number, say revenue. Select the revenue field, drag and drop the same on to the “TEXT” area of “MARKS” pane for editing the same. Click on the field and select option. Use the formatting to change the number type to currency and display units to millions and change to dollars. Also, let me apologize for BAD Screenshot Quality if any.
Suppose you want to select a dimensions variable say Name and want to identify the total number, the method is to drag and drop the variable on to TEXT in the MARKS pane and change the attribute from Dimension to Measures –> count distinct. Count distinct since there are repetitions in the variable and it only sane to select the same, as there will be repeat customers. The COUNT will be relevant for a variable like ORDER ID which should be unique from a sales perspective.
Suppose we don’t know where sales data is from (which would be LOL moment for an analyst), we can use the geographic indicator that Tableau generates.
Formatting maps is easy which is available in the Taskbar. Observe above the combinations in the panes to get the desired visualization. Believe me, an ambitious beginner/learner must play around nonsensically with the variables to derive meaning to the experiment and gain the experience.
Some Insight Manipulations — People Love Top Ten
Revenue cities or regions - Add the variable to Filters. Follow the prompt “TOP” for the purpose and select by field and create the filter.
Visualizers — The essential learning for Tableau is that you Run Trials and explore to understand the differences!!!
Suppose you want to explore the revenue per region on subsegment of wines, the variable combinations are that of REVENUE, REGION and SUBSEGMENT. A good convention is to make a table, 2D data, with row and columns, correspondingly the DIMENSION variables, and datapoints as values of the third variable REVENUE. The data table formed can be depicted using some charts in the SHOW ME tab on the top-right end. Here the requirement is to incorporate 3 variables, so the choices of normal PIE charts (extremely useful to show visual impact of Percentages/Share), BAR/COLUMN, Scatter, Histograms etc., won’t work.
Box Plot can be a powerful three variable chart (at least ONE Measure needed). However, here the best visualization alternative to PIECHART would be TREEMAP. Visually, the map is powerful and clearly indicates to anyone that the South region hold the key to the company’s pockets. Long Live the Drinkers…. LOL!!!
Power of Boxes
A Box Plot is the trump card of any impressive visualization. It not just impresses the clients on the analysts’ skills but connects with them easily too.
For descriptive analysis let us use Reference Line from Analytics and DnD(drag-and-drop), to understand the statistics corresponding to the analysis. Also, selecting the median instead of average as median is more important to counter outlier effects. The business problem could be “Analyse the affordability of the Wine Brands”. An analyst can also assess using a threshold instead of median.
Here, what do you think is the price point (average Unit price) for a user?
Univariate is a variable’s introspection.
“In Australia’s Barossa Valley, winemaker Victor was puzzled by the sales pattern. Entered Lucia, the Numbers Wizard, from the Ozzie School of Statistical Wizardry. Her Univariate Spell revealed that sales of one of his wines, Shira, peaked in winter, especially during festivities. She gave Victor the spell. Victor marketed his Shira as a winter treat and was ever indebted to Lucia. Lucia shares a key insight: sometimes, understanding just one variable can chart the path to success.”
An example of a univariate analysis would be Histogram, and it counts the FREQUENCY of any variable analyzed. So, the second variable in the axis should be the COUNT (variable). Of course, taking unique count also doesn’t make any sense here.
Chart says that 99% of the price lies within 200($), which concurs to the box plot values. Observe the other values and compare it to the box plot values. The price distribution is right-skewed. Psst… It is also better to Convert the measure from discrete to continuous (interval based) for histogram, i.e., 10–20,20–30 etc.
Try newer things and Publish... Publish... Publish…
Visualization link: http://rb.gy/5omwn