Graphing of Complex Data

Quick reference

• 00:05 Hi, I'm Ray Sheen.
• 00:06 We've looked at some basic graphical analysis techniques.
• 00:10 Let's consider now a few that work better with more complex data.
• 00:15 I'll start with putting data into tables.
• 00:17 I know it's not really a graph, but there are a few guidelines to follow
• 00:21 to make it easier to find and understand tabular data.
• 00:24 Sometimes there are multiple correlation relationships
• 00:27 across attributes within a data set.
• 00:29 And these are often hidden when the data is in a table because all the numbers
• 00:34 just kinda blend together.
• 00:35 There's no picture of what the data is saying.
• 00:38 However, when a data point does have multiple attribute values,
• 00:42 I like to start with a data table.
• 00:44 In the table each data point is a row, and
• 00:47 the column represents the different attribute categories,
• 00:50 like in the table shown, where we have associates in a call center.
• 00:55 And we have their name, their years of experience, the number of errors or
• 01:00 defects that were created in the past month, and
• 01:03 their professional certification level.
• 01:05 Now to understand what the data is telling me,
• 01:08 I can sort the data based upon any column.
• 01:10 For instance, I could sort the table based upon an alphabetical listing of the names,
• 01:14 although that probably would not provide any useful insight.
• 01:18 However, when sorting by the number of years or experiences shown here,
• 01:22 we can draw a few conclusions.
• 01:24 We see that the number of years experience does appear to
• 01:27 correlate with the training and certification level, but
• 01:29 it does not appear to have any significant effect on the number of defects.
• 01:33 Apparently, neither experience or training and
• 01:36 certification seem to affect that number.
• 01:39 One suggestion when using tables for showing data,
• 01:42 try to use units that result in only 2 or 3 significant digits.
• 01:46 Don't show data tables with long strings of numbers.
• 01:50 They become mind-numbing for most stakeholders.
• 01:53 Let's switch now to the horizontal bar chart.
• 01:56 I've already discussed the vertical bar chart, and
• 01:59 the horizontal bar chart has many similarities.
• 02:02 It is an excellent chart for quantifying instances of an attribute and
• 02:05 showing that graphically.
• 02:07 The length of the horizontal bar shows the magnitude of the attribute being measured.
• 02:12 To make it easier to read,
• 02:14 I recommend that you sort the categories from largest to smallest.
• 02:17 That way, the most significant items appear at the top of the chart.
• 02:21 One of the advantages of the horizontal bar chart over the vertical bar chart is
• 02:25 that you can exceed the eight category rule that we use with vertical charts.
• 02:30 You can have as many rows as you want.
• 02:33 I personally have used over 50 rows, and
• 02:35 I have seen horizontal bar charts with even more than that.
• 02:39 On a vertical bar chart,
• 02:40 it becomes hard to read categories when there's a large number.
• 02:43 That's not the case on a horizontal bar chart.
• 02:46 However, like with the vertical bar chart, I recommend that you use just one color.
• 02:51 I've occasionally used multiple colors so
• 02:53 that it was easier to distinguish the lines.
• 02:55 But in that case, I always repeated the same pattern for the colors so
• 02:59 that no one would start to think that different colors meant different things.
• 03:04 One caution when using the horizontal bar chart,
• 03:06 don't use it with a time-based attribute.
• 03:09 Inevitably people will start to look at the horizontal scale like it
• 03:14 is a time scale instead of a count of the number of instances.
• 03:18 Next I wanna discuss the pie chart.
• 03:20 What makes this chart a little more difficult is it is showing percentages
• 03:24 instead of the raw data.
• 03:26 Now you can build it with raw data, but the visual representation
• 03:29 of the categories is always as a percentage of the whole.
• 03:33 This chart is really good for doing comparison charting,
• 03:36 showing how the percentage of a category changes based upon different conditions,
• 03:41 such as location or product line.
• 03:43 Or comparing the before and
• 03:44 after condition of a process that you are analyzing.
• 03:48 To make it easy to read, I try to limit to about 6 slices of the pie.
• 03:53 You can go a little higher if you have roughly equal slices.
• 03:57 But if you have dominant slices,
• 03:58 the minor ones are miniscule by the time you get over 6 slices.
• 04:03 And in this case, colors are very helpful for
• 04:05 being able to discriminate between the slices.
• 04:08 As with the vertical and horizontal bar charts,
• 04:11 this chart helps to show why factors are dominant.
• 04:14 Of course, the big difference is that in this case you're using percentages.
• 04:19 The last chart to discuss is the box plot.
• 04:22 Box plots are ideal for showing comparison between data sets because
• 04:27 it shows the median and extremes of the data sets and a level of central tendency.
• 04:33 I use box plots with attribute data to compare how data sets for different
• 04:37 locations or customers or product lines look with respect to each other.
• 04:42 The box plot quickly reveals when the data set for
• 04:45 one attribute item is very different from all the others.
• 04:49 When discussing box plots, we often talk about the box and whiskers.
• 04:53 As you can see, each data set has a box with a vertical line somewhere inside it
• 04:58 and two horizontal lines extending to either end.
• 05:01 To create the box and whiskers, five values are needed from the data set.
• 05:06 To get these five values,
• 05:07 first order the data within the set from the largest to the smallest.
• 05:11 The largest and smallest values are the two endpoints for
• 05:15 the horizontal lines, the whiskers.
• 05:17 The midpoint of the data, or median, is the vertical line that is within the box.
• 05:23 The 25% and 75% points of the data are the vertical sides of the box.
• 05:30 This plot then shows the data spread and central tendency of the data and
• 05:34 whether the data is biased towards the high end or the low end of the spread.
• 05:38 Comparing boxes for different data sets
• 05:41 can usually reveal which ones are the same and which are different.
• 05:45 In the illustration shown here, the four boxes have a similar upper value.
• 05:50 However, the lower value, midpoint, and width of the boxes are quite different.
• 05:55 The top two boxes are similar, and the bottom two boxes are also similar.
• 06:00 These additional graphical techniques can be very useful when your data sets
• 06:05 have several levels of complexity.

Lesson notes are only available for subscribers.