Basic Graphical Analysis

Quick reference

• 00:05 Hi, I'm Ray Sheen.
• 00:06 The analysis of data does not need to be a statistical analysis,
• 00:10 many times a graphical analysis will provide very helpful insights.
• 00:15 A graphical analysis means that we chart the data creating
• 00:19 a picture of the data set.
• 00:21 Now during the Measure phase, the Lean Six Sigma team is collecting data,
• 00:25 often a great deal of data.
• 00:27 If that data is just tables of lists of numbers, it's hard to make sense of it.
• 00:32 A graphical display is usually much better for
• 00:34 communicating the data than mesmerizing tables of numbers.
• 00:38 There is an old saying that a picture is worth a thousand words, and
• 00:41 that is definitely true with Lean Six Sigma data sets.
• 00:45 The picture puts the data in context, in particular if you use the appropriate
• 00:50 charting technique, the picture of the data will often clearly show the problem.
• 00:55 Either the root cause or the process effect.
• 00:59 Each graph or chart is well suited to a particular kind of data.
• 01:03 Let's look at several of the most common charting approaches and how to use them.
• 01:07 I'll start with the vertical bar chart, which is also called a histogram.
• 01:12 This chart is the most common and basic chart.
• 01:15 We see them all the time, and that's because it is such a simple, but
• 01:19 powerful chart.
• 01:21 In particular it is well suited for attribute data.
• 01:25 You just make each of the columns the categories of the attribute,
• 01:28 and then count the instances.
• 01:31 It then shows the magnitude of each of those categories.
• 01:34 A good rule of thumb, in order to make the chart easy to read,
• 01:37 is to limit the number of categories to just eight categories.
• 01:41 If you have more than eight,
• 01:42 accumulate the minor categories into one miscellaneous category.
• 01:46 We normally want to use this chart to highlight which categories are large,
• 01:51 and therefore, important, and which are not.
• 01:54 For aesthetics, it's usually better to keep the bars the same color.
• 01:58 The observer then focuses on the height of the bar and not the color.
• 02:02 The exception is if you're using the chart in a before and
• 02:06 after view of the categories, in that case you can use different colors for
• 02:11 the two different cases.
• 02:13 If this chart looks familiar it's because we have already used it when we discussed
• 02:17 the Pareto principle, a Pareto chart is a special case of the vertical bar chart.
• 02:24 The next type of chart is called a line graph, this is also a simple graph to
• 02:29 generate it shows you relationship between two process parameters or
• 02:33 factors, and this type of chart works really well when at least one of those
• 02:37 factors is variable data, in other words that factor can take on any value.
• 02:43 The horizontal scale, which represents one of the factors,
• 02:46 normally occurs in set increments, in a sequence.
• 02:50 The most common factor is time, although,
• 02:52 it could be other factors like temperature or dimensions.
• 02:56 The vertical scale represents the value for the other factor.
• 03:00 The data points are then connected by a line.
• 03:03 This is to indicate that there is a continuity of relationship.
• 03:06 Now often the line will reveal patterns or high point or a low point.
• 03:11 This pattern or inflection point is typically significant in understanding the data set.
• 03:16 Just like the Pareto chart is a special case of the vertical bar chart,
• 03:20 the run chart is a special case of the line graph.
• 03:23 Now we'll use the run chart in more detail in another module of this program.
• 03:28 The third basic graphical analysis is the scatter diagram.
• 03:32 This works well when showing the relationship between two numeric
• 03:36 factors or process parameters.
• 03:38 That means it's especially well suited for
• 03:41 the case where both factors are variable data.
• 03:44 Now the purpose of this chart is to determine if there's a correlation between
• 03:48 the two factors.
• 03:50 When a correlation exists, the pattern of data begins to look like a line or
• 03:55 an ellipse that has a slope to it.
• 03:58 The tighter the ellipse, and
• 03:59 by that I mean the more it looks like a line, the closer the relationship.
• 04:04 And the closer the slope is to 45 degrees, either positive or
• 04:08 negative, the tighter the correlation.
• 04:12 If there is no correlation, the plot of the points will be totally random,
• 04:17 showing that one factor does not impact the other factor.
• 04:20 Or the plot could be a horizontal or vertical line.
• 04:24 Again showing that regardless of how one factor changes
• 04:27 the other factor remains constant.
• 04:29 One caution when using this graphical technique.
• 04:32 Don't assume that because there's a correlation, there is a causation.
• 04:37 I may spend a lot of time in a shopping mall and spend a lot of money.
• 04:42 But the reason for both of those is because I wanna buy a lot of shoes - uh, not.
• 04:49 Graphical techniques are a great way to communicate what's important with the data
• 04:53 that's been collected.
• 04:55 You don't have to use statistics if the graph already tells the story.

Lesson notes are only available for subscribers.