D
M
A I
C
Tool description courtesy of © Processfix Limited 2008
Histogram
Description
Analysing data provides a great insight into the behaviour of a process, but when there is
lots of it, a picture or graph is almost always easier to analyse than a set of numbers.
The histogram is a graph that, once constructed, helps to identify whether the process is
capable of meeting the customer’s expectation and provides an insight in to its behaviour. It
provides a means to visualise the frequency (occurrence) of data, and thereby allows
interpretation of process performance by analysing the centre of the data, the amount of
variation and any significant changes or anomalies.
The shape of the histogram gives further insight in to the behaviour of the process. Shapes
that differ from the normal distribution (a symmetrical bell curve) indicate that other factors
may be affecting the process, such as imposed targets, constraints or perhaps secondary
processes.
Where to use
•
To assess the level
of variation in a process
•
To assess anomalies in the distribution
•
To uncover the root cause of process performance
•
To compare multiple sets of data, for example before and after improvement
•
To investigate process performance in relative to the customer expectation
How to use
1. Implement a data collection plan to collect the measures required to be analysed
2. Collect at least 50 data points to be confident that the sample will generate a
representative histogram and distribution
3. Draw the horizontal (x) axis so that it encompasses the full range of data
4. Group the data into 10-20 intervals across the full range of data (depending on the
sample size and resolution of the histogram required)
5. Count the number of data points that fall within each interval and plot the count
(frequency or occurrence) on the vertical (y) axis
6. Calculate and mark the average of the data on the histogram
7. Identify the customer expectation and mark this on the graph with a vertical line
8. Compare the histogram to the normal distribution (bell curved and symmetrical) to
understand how the process behaves and to identify anomalies
9. Assess the centre, spread and shape of the distribution to uncover the likely causes of
variation
10. View the areas of the graph that fall outside the expectation limit(s) as the opportunity
for improvement.
D
M
A I
C
Tool description courtesy of © Processfix Limited 2008
Handy tips
Just because the average meets or exceeds the customer expectation does not mean that
the process is working well.
View the spread outside the expectation limit as the opportunity for improvement.
Compare the histogram to the normal distribution (bell curved and symmetrical) to identify
anomalies in process behaviour.
Where there is more than one obvious hump, this signifies one or more processes in
operation, for example a right-first-time process and a re-work process
Always compare the histogram to local knowledge of the process. Knowledge of the data
itself does not provide insight into potential improvements.
Be wary of histograms that have very few columns as this may indicate a small sample size
or low resolution
of the measurement system
Plot the same data in run chart form in order to detect trends and special cause in process
performance.
Example application
Central government requires local planning authorities to make their planning decisions
within a target time period. The planning authority’s ability to meet this target has a direct
impact on their central funding.
Constructing a histogram with number of days to process the application on the horizontal
axis and the frequency of applications on
the vertical axis, the performance of the process
was plotted for applications received in the previous 12 months.
With a range of 0 to 90 days, the horizontal axis was broken in to intervals of 5 days and the
number of applications plotted in each interval. The results showed a prominent spike before
the target time with most applications being completed just before the target date.
The spike suggested that there was a frantic push at the end of the process to attempt to
meet the target date. This suspicion was confirmed by those in the process who considered
that application with a quick turnaround time were not a priority and thus delayed, often to
the last moment in preference for those close to the deadline for a decision.
The reason for this was the need to meet the target deadline rather than a drive to issue
decisions in a timely manner. This target centric thinking became self defeating in that it
caused a constant panic to return planning decisions before the looming deadline.
D
M
A I
C
Tool description courtesy of © Processfix Limited 2008
Consider using with
•
Value stream map
•
Run
chart and histogram
•
Probability
plot
•
Affinity
diagram
•
Fishbone
diagram
Facilitation time
30 mins
D
M
Number of BARs Customer specification limit
A I
C
Tool description courtesy of © Processfix Limited 2008