Deep dive – what insights can the dashboard reveal?

I expect that some stocks may be more volatile than others and some may grow better than others. Please note that, as I am writing this, I do not know what insights I am going to uncover ahead of time—you are following me live as I use the power of Spotfire to find fresh insights in the data:

1. Try clicking on some individual KPI tiles—notice that the line chart will update to show details for the data that corresponds to the KPI that has been clicked. The line chart shows the detailed trend of the KPI over time.
2. What if you want to view multiple KPI tiles at the same time? Most Spotfire visualizations support click-and-drag marking, where you mark an area on the chart to show the details underneath the area that's been marked. The KPI chart works slightly differently.

Selecting multiple KPI tiles will allow you to compare them together. Here, I have selected Consumer Discretionary and Consumer Staples:

Notice that the two trends are shown side by side in different colors on the line chart. From the line chart, the following insights are clear:

• The value of Consumer Discretionary is always higher than Consumer Staples
• The trend of Consumer Discretionary is less stable than Consumer Staples—there's a big dip around January-March 2016

These insights might lead to a direct action from this chart alone! For example:

• An investor that wishes to make short-term, larger gains may choose to invest in unit stocks in Consumer Discretionary
• A more cautious investor, for example, one heading toward retirement, may choose to invest in stocks in Consumer Staples

We can take this analysis even further by drilling in to the makeup of the individual industry sectors. Right now, the line chart shows the difference between various KPI tiles, but does not yet provide any insight into why one industry sector might be trending differently from another.

Let's enable the splitting of the data by stock as well as sector by creating another line chart and splitting it by stock name. We can duplicate the line chart so that we can create another from scratch:

1. Right-click on the line chart and choose Duplicate Visualization.
2. Now, split the right-hand chart by name as well as sector—choose Name from the Line by selector:

Interesting! Look at the screenshot of the second line chart now:

It's pretty clear that some Discretionary stocks are indeed generally higher than the Staples stocks—look at how the Staples stocks are all bunched at the bottom of the visualization. Notice also that there is one stock that is head and shoulders above the rest—both in terms of its overall value and in the magnitude of its growth. Right now, we don't know which stock that is, but it can be found easily enough.

Simply hover the mouse over the line chart and particularly, the line at the top:

So, the top performing stock is PCLN—this is the Priceline Group (from 2012 to 2017 it has trebled in price). It should be far from me to give financial advice in a Spotfire book, but... you can draw your own conclusions!

I'd really like to remind you of one other nice feature. In Chapter 2, It's All About the Data, the concept of natural hierarchies in data was introduced. Time/date data is one of the most interesting and useful natural hierarchies there is, and here we have used it for the first time. Think about it—time and date information is split up into this kind of hierarchy:

• Year
• Month
• Day
• Hour
• Minute
• Second
• Millisecond

However, it could also be split up as follows:

• Year
• Quarter
• Month
• Day

In addition, there are concepts of fiscal years, quarters, months, and so on.

Spotfire has a native understanding of such date/time hierarchies, and we can use that to our advantage by following these steps:

1. Drop-down the x-axis selector on the second line chart and choose:
   Date, then Year >> Quarter >> Month >> Day of Month:

1. Now, adjust the hierarchy slider on the axis. Try all the different levels of the hierarchy in turn and watch what happens to the line chart. Spotfire will automatically aggregate the line chart at each level. Here is the highest level (Year):

The next level is Quarter:

Now Month:

And finally, Day of Month:

Notice how each line is aggregated at the level currently selected in the hierarchy—the line is smoother at the less detailed levels and more jagged at the more detailed. Higher levels of the hierarchy are useful for looking at overall trends and minimizing local effects (rapid changes that don't affect the overall statistical significance of the trend too much). Lower levels are useful for examining the detail and looking at variations—peaks and troughs over time.

Finally, let's hone in on some detail and try to understand more about why there is a dip and a rebound in the value of Consumer Discretionary stocks at around January 2016. Was it caused by PCLN on its own or was there a sector-wide issue leading to a crash and a rebound?

How can we look at this area of the visualization in more detail? Spotfire's zoom sliders will allow us to do this really easily.

If you are using a Spotfire Analyst client, perform these steps:

1. Right-click on the visualization and choose Visualization Features | X Zoom Slider:

1. Do the same again for the Y Zoom Slider.

In web clients, perform the following steps:

1. Right-click on the x-axis of the line chart and choose Properties.
2. Check the Show zoom slider checkbox:

1. Close the Properties dialog.
2. Do the same with the y-axis and enable its zoom slider.

Now, you can experiment with the zoom sliders to explore various features of the data in detail. It really does look like PCLN's dip and rebound is confined to January to March 2016. I wonder what happened during that time:

In fact, I think the dip and rebound isn't big enough to affect the overall average that much—remember that the same trend was found overall for this industry sector. We can easily test this hypothesis by removing the effects of PCLN from the analysis:

1. Click on the line representing PCLN. This will mark it.
2. Right-click on the visualization and choose Marked Rows | Filter out:

1. Now, look at the line chart on the left. The dip and rebound is still there! So it can't be due to PCLN alone:

That makes me think. Could it be that, in fact, Consumer Staples is the industry sector that bucks the trend? It might be... let's just check:

1. Go back to the KPI chart and select some more industry sectors—I have chosen the following:
   • Consumer Discretionary
   • Consumer Staples
   • Energy
   • Financials
2. Look at the line chart trends now:

It would seem that the only sector unaffected at that time (in this subgroup anyway) is Consumer Staples. However, notice that it has a dip and rebound a little bit later than the other industry sectors.

I have also added labels to each line—the setting for this is available in the visualization properties—go and take a look and try turning them on for yourself.

The updated line chart suggests to me that there must have been a problem with the stock market as a whole in January-March 2016. In fact—googling "stock market crash January 2016" does point to a bit of a crash and rebound in the first quarter of 2016—see how our analysis enabled us to identify this and explore the details of it?

If I were doing this analysis again and knowing what I know now, I may well have just produced a single line chart with the average of all stocks over time—I leave it as an exercise for you to do if you so wish!

In summary, I know that it may seem that focusing on PCLN may have been a "red herring"—that is, of no real value, and that my initial assumptions about the markets (that in some way, Consumer Discretionary was different from the other sectors, and that PCLN was leading to this difference) were wrong.

However, doing this kind of exploratory analysis is a very useful technique when working with analytics. I have also enjoyed showing you how to drill in to details and configure Spotfire visualizations "live" without being worried about designing them ahead of time. How can you know how to design something if you have no idea what you need to design ahead of time?

Having a hunch and following it, only to be proved wrong, is just as valuable as having a correct assumption in the first place. Go with your instincts. Investigate the data. See what it reveals!