Locked lesson.

Upgrade

About this lesson

Visual analysis techniques are particularly good for illustrating significance, similarities or differences, and correlation between parameters within a data set. Visual data may provide insights that are hidden in statistical analysis

Quick reference

Basic Visual Analysis

Graphical analysis techniques are particularly good for illustrating significance, similarities or differences, and correlation between parameters within a data set. Visual data may provide insights that are hidden in statistical analysis

When to use

The techniques are used in the Analyze phase of the project. The graphical analysis techniques work very well when there is either a special cause problem or a singular root cause for the problem. In those cases, it is often faster to use the graphical visual analysis than a statistical analysis. Also, it is much easier to explain the problem to stakeholders and team members using visual analysis than with statistical analysis.

Instructions

Visual graphical analysis is used to provide insight into a problem. The “picture” presented in the graph is often a better tool for communicating with stakeholders than providing a detailed statistical analysis. The visual analysis can provide context. In addition, when using the correct analysis technique for the situation, it will highlight the presence of a problem or the lack of a problem.

There is a problem with only using averages when describing a dataset. While the average is representative of the most common experience, it does not highlight the impact of outliers. However, a customer’s experience is based on their unique instance of the process or product and if they experience the outlier, they could be very pleased or very displeased. For that reason, the nature of the distribution is also important. In fact, the four diagrams below have the same number of data points and identical descriptive statistics, but they represent very different “real world” situations.

The first dataset represents the condition of positive linear correlation. It is reasonably strong, but there is other variation also occurring. The second dataset has very strong non-linear correlation and within the data there is a local maximum. The third dataset has strong linear correlation with a single outlier. The fourth dataset has no correlation with a single outlier. These represent four very different real-world conditions that are evident from the visual analysis but are not evident from the statistical analysis.

Hints & tips

  • Once the data is captured in a database or table, it is easy to create charts and graphs in both Excel and Minitab.  It only takes a few mouse clicks, so visualize the data first before jumping into statistical analysis.
  • Many stakeholders have no statistical training or skills, but won’t admit that they don’t. A visual analysis is a better way to ensure effective communication of issues.
Login to download
  • 00:05 Hello, I'm Ray Sheen.
  • 00:06 Many times the fastest and easiest type of analysis that you can do in
  • 00:11 a Lean Six Sigma project is a visual analysis.
  • 00:14 I've already discussed many of the types of charts and
  • 00:17 graphs that you can use to display data.
  • 00:20 Let's take a few moments and review them from the standpoint of data analysis.
  • 00:25 Keep in mind that graphical displays of data are excellent ways to communicate
  • 00:29 the data to both stakeholders and other team members.
  • 00:33 I suggest you introduce a graphical display of the data early in a problem
  • 00:37 analysis session with the team.
  • 00:39 Often the underlying causes will jump out from the picture of the data, and
  • 00:44 the team knows where to focus any further analysis.
  • 00:48 The visual display does this by putting the different data elements into the same
  • 00:52 frame of reference or context with each other.
  • 00:55 Then differences and similarities are easy to spot.
  • 00:58 The picture draws the eye to the point of change or difference.
  • 01:02 Or if there is no significant difference,
  • 01:05 the picture illustrates that the data is the same.
  • 01:09 Now, different types of displays are best for different types of analysis.
  • 01:13 We discussed this in an earlier lesson, and
  • 01:15 we'll deal with it even more in the next one.
  • 01:18 Some graphs highlight what's dominant in a dataset while others can bring in elements
  • 01:23 of the process, such as time to see when a process results suddenly changes or
  • 01:28 has it been a gradual change?
  • 01:31 I want to highlight the need for
  • 01:33 visualization first by focusing just on averages or mean values.
  • 01:38 We often work with mean values in the data analysis as we start to compare data
  • 01:43 sets to try to find the cause of the problem we're studying.
  • 01:47 But averages alone do not tell the whole story.
  • 01:51 Customers don't feel the average, they feel their unique instance of the process.
  • 01:56 And it's a good bet that the customer who feels the poor performance out at
  • 02:00 the extreme of the dataset will be the one who yells the loudest.
  • 02:03 That's not to say that the average might not be a problem also.
  • 02:07 I worked with a call center one time where the average hold time for
  • 02:10 customers calling in was a little over 30 minutes.
  • 02:13 Actually, that was the average hold time for those who would sit around and
  • 02:17 wait for the 30 minutes.
  • 02:19 Lots of customers just hung up before they ever got answered.
  • 02:22 Needless to say, this call center had very poor customer reviews.
  • 02:27 In fact, in the example I'm showing here, two call centers have identical,
  • 02:31 average wait times of 7 minutes.
  • 02:33 The call center on the left has people holding 5 to 10 minutes all the time.
  • 02:38 To improve the average, they'll need to change the system,
  • 02:41 this system is stable and it's stable at that level of about seven minutes.
  • 02:45 The one on the right has three customers on hold for just a minute and
  • 02:49 one on hold for 25 minutes.
  • 02:52 There is an obvious special cause occurring for that one customer.
  • 02:56 If we find that root cause of prevent it, we can have outstanding performance
  • 03:00 without having to otherwise change the process.
  • 03:04 Let me give you another illustration, which is why it's so
  • 03:06 important to graph your data.
  • 03:08 Raw data, even statistical analysis of raw data, can be misleading.
  • 03:15 Data visualization, which means translating the data points into a graph
  • 03:18 or a chart, can really help the process to reveal what's happening.
  • 03:22 I have four different datasets, and
  • 03:24 first I did a statistical analysis on these datasets.
  • 03:28 Every dataset had identical mean values and variance.
  • 03:32 The correlation between the two parameters was the same in each data set, and
  • 03:36 that correlation was high, showing a strong relationship.
  • 03:40 If we took the data points and graphed a straight line through them.
  • 03:43 All four data sets had the same slope and the same y-intercept.
  • 03:49 So statistically, we might want to say that all four of these are virtually
  • 03:54 identical, but the statistical dataset doesn't tell the whole story.
  • 03:59 Look at the four graphs With the data, they are very different.
  • 04:04 The first one on the left is a typical II parameter positive linear correlation.
  • 04:10 The next one is very obviously a nonlinear correlation of the parameters.
  • 04:15 They are tightly related but they are nonlinear and
  • 04:18 in fact we have a clear maximum point within the data.
  • 04:22 The third one looks like a strong linear correlation with a special cause
  • 04:27 occurring at one point.
  • 04:30 Another possibility is that there was an error in recording the data at that point.
  • 04:34 Either way, that point needs to be investigated.
  • 04:38 The last one on the right is very interesting I would say there's probably
  • 04:42 no correlation relationship between the two variables.
  • 04:46 Rather, there was a data entry error or
  • 04:49 in some other way a data point was contaminated.
  • 04:53 These four different conclusions are obvious from the graphs, but
  • 04:57 they are not obvious from the statistics of the data sets.
  • 05:00 >> Don't overlook the benefit of visual analysis, many of
  • 05:04 the basic problem solving methodologies rely solely on visual charts.
  • 05:09 They're especially good for finding special cause problems.

Lesson notes are only available for subscribers.

PMI, PMP, CAPM and PMBOK are registered marks of the Project Management Institute, Inc.

Gift this course