Run Chart

Viewing 7 posts - 1 through 7 (of 7 total)
  • Author
  • #33537

    Simpy Sadana

    How and why do we make a Run chart and what is its significance?


    Ex Ex MBB

    There are several tools that you can use to evaluate process performance over time:

    Run charts: (ala Minitab) view this as a naked control chart.  You have a center line (which you can place as the mean or median), but you do not have control limits.  Minitab lets you do several things: you can actually subgroup if your samples have employed a rational subgrouping strategy; there are also statistical tests (for trends, oscillations, mixtures, and clustering); and of course, you can “eyeball” the chart to see if there are any patterns that would suggest trends, oscillations, mixtures, and clusters.
    Control charts: lets you assess stability of the process over time.  Will actually generate signals for out of control conditions.  You can also subgroup data.
    Time Series Plot: just a plot of the data over time.  Strictly an “eyeball” chart.  To the best of my knowledge, you cannot automatically subgroup the data in Minitab
    Hope this is what you were seeking…



    full deatial on run chart.


    fake accrington alert

    You  shouldn’t  ask  such  basic  simple  questions



    Hi All:
    I have been in this field (six sigma) for around two years now, am still learning about the methodology & tools. Would really appreciate if all the knowledgeable people on this forum can help me with my questions.
    Just saw this thread on Run chart, so wanted to clarify my doubts as well:
    In Minitab the chart gives me information about four types of non-random behavior in my data, viz.
    1. Cluster, 2. Mixtures, 3. Trends, & 4. Oscillations
    My questions are:
    a. If I don’t look at the p value, how can I by just looking at the chart say that there is any one of the four non-random behavior in my data? How will a clustered data or mixture or trend or oscillation look like?
    b. Is  there any rule / thumb rule about the number of data points to be used for run chart? If I am working on weekly data, how many weeks of data will I require. 
    Is there any link from where I can get all this information. I have searched on google but nowhere did I get the detailed information.
    Any help is highly appreciated.



    Davis Balestracci has a good article on Run Charts in Quality Digest this month:



    Clustering:  lot of data points fall on one side of the median (next to each other) and indicates periodic shifts
    Mixture:  Absence of points near center line with 14 or more data points alternating up or down
    Oscillation:  Data points constantly fluctuating  up and down rapidly around median
    Trend:  7 or more data points continously increase/ decrease

Viewing 7 posts - 1 through 7 (of 7 total)

The forum ‘General’ is closed to new topics and replies.