Spot Trends and Focus Resources With Data Visualization
Gary Goldberg of Information Builders and Doug Mack of IBM Systems Lab Services explain how data visualizations allow analysts to quickly spot trends.
By Doug Mack and Gary Goldberg02/01/2019
The idiom “a picture is worth a thousand words” suggests that you can derive a lot of information in a shorter amount of time from a single picture than you could from digging through a lot of text. The same could be true for data analysis—a visual depiction of data is worth a thousand spreadsheets.
A powerful trend lately in the business intelligence (BI)/analytics space is the use of visual depictions of data as a starting point to easily spot trends, anomalies or exceptions in large amounts of data. Imagine the alternative of staring at a ton of data in spreadsheets.
The Problem With Spreadsheets
An IBM i client recently asked how he could get data from Db2* for i into multiple sheets of a spreadsheet, and then use Excel functions to build charts and graphs over that data in multiple sheets. For instance, perhaps they wanted to look at product mix metrics such as gross profit or quantity sold by product mixes, and by a time period, such as year, with each of those time periods in different tabs of the spreadsheet. Each year’s worth of data would be in a separate tab of the sheet, with charts built over each set of yearly data.
The approach is doable, but may not be optimal for analysis. At the end of the day, you’re left with massive amounts of data in spreadsheets. What’s more, data in spreadsheets is editable, so sharing them introduces data integrity (and security) issues.
Interact With Your Data
A much more effective approach would be an interactive dashboard over secured data that could show you in a very visual manner the metrics of interest across all of product sets and time periods and allow you to interact with that data without the need for data extractions into spreadsheets.
Figure 1, for example, contains a lot of information on a single view. The Quantity Sold by Product subcategory is in the top heat map, and gross profit across product categories by year appear in a matrix line chart below. Being able to quickly identify interesting information by colors alone allows a business analyst to determine what to dig deeper into.
As mentioned, this isn’t a static view of data. From here, you can interact with the visual depiction of data in many ways. With one click of the mouse, you can see the data behind the charts. Hover over a section of interest and you see the pop-up menu allowing you to filter or exclude data. You can also drill down to a lower level of data, such as the month of the year to the day of the year. You can even drill up to look at the metrics by quarter. Slider bars, check boxes and other controls provide even more ability to manipulate the view.
Location, Location, Location
Another interesting trend is location analysis. Location analytics is the notion of analyzing data with a geographical perspective. For example, location analytics allows you to determine:
- What are your sales versus the weather forecast or gross profit of your product set from a regional perspective?
- How can you represent that by attaching those metrics to a map representing each of your regions or store locations?
You can more easily spot regional patterns, trends and exceptions, and then focus on that region with interactive filtering and drill downs. Dynamic interaction with a geographical representation of data can be a very powerful way for the data analyst to diagnose what’s happening in the business and create change.
In Figure 2, you can easily spot individual states of the U.S. with higher (and lower) levels of overall revenue by aligning the revenue measurement to a specific state. Note the geographic location could be country or city, zip code or latitude/longitude associated with your business locations.
Layer Data for Additional Insights
Location analytics becomes even more interesting with the ability to add layers of information on top of the map.
A good example is the ability to include demographics, such as population or average income, in your operational data analysis. Suppose you could easily recognize that a product isn’t selling well in a particular part of the region. Your first instinct might be to drop that product from that region. But with the ability to see other metrics in the same geographical view, you determine that this is a highly profitable product, and that region’s growth rate is one of the highest in the geographies you sell into. The insight you glean from such a visualization would suggest the exact opposite action should be taken.
If you’d like to see more of these visualization examples, visit ibm.biz/db2wq-221-videos.