A Histogram is a bar chart showing the frequency of an outcome. In Six Sigma, we can use a histogram to visualize what is going on. A Histogram can reflect the voice of the process.

How to Use a Histogram

Six Sigma practitioners can use the pattern reflected in the histogram to discern a process variation. This is an kind of data visualization. Here, a histogram is useful in evaluating the shape of the data.

Some of the questions we want to evaluate when interpreting a histogram are; How many peaks are there? Are there outliers? does it look approximately symmetrical and bell-shaped? If not, the process being reflected ins not a  normal distribution.

Histogram shapes
Histogram shapes

Bell-Shaped Histogram

If there is a bell shape, your data is normally distributed and hence, no variation (or influence from other factors like the 6Ms).

Histogram with Spikes

If there are multiple spikes in the chart, there is likely variation in the process.

Even Histogram

On the other hand, if all of the bars in the histogram are at the same level it’s not likely that we are measuring the process in the correct manner.

Doing statistical analysis usually involves the following:

  1. Measuring of the central tendency.
  2. Measure of the variation.

Creating a Histogram:

1) Get Data

Ex. Delay times in mins for an airline

Use continuous data– data collection from a frequency distribution check sheet.

Histogram data set

Step 2: Order it and Assign Categories

Histogram ordered categories
  • Note: The number of cells (columns) has an effect on the shape – so choose wisely!
  • Could use 2^k = N where k = # of categories (smallest positive integer) and N = # of data.
Number of categories to use in a histogram depends on the sample size.
Number of categories to use in a histogram depends on the sample size.

Step 3: Create a Bar Chart Preserving Counts and Categories

histogram occurrences per category

How to Create a Histogram in MiniTab

 Notes on Histograms

  • While a normal histogram just tells you the data, a well-constructed histogram tells you more than just the raw data
    • A normal histogram should help see:
    •  A well-constructed histogram should help you:
      • Visualize the voice of the process using the raw histogram itself.
      • Be easily comparable to the voice of the customer.
      • You could even include the specification limits to reflect customer expectations.
      • A picture is worth a thousand words.
  • Different from frequency distributions:
    • A Histogram is a bar graph version of a frequency diagram
    • Frequency distribution represents data with x’s or check marks.
    • A continuous relative frequency graph is called a histogram.
  • Set up a histogram using the appropriate number of cell intervals
    • # of cells can affect the shape of the curve
      • Might hide distributions if too broad

Histogram Videos

Comments (4)

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.