Using Mode of Data as Data Point
Six Sigma – iSixSigma › Forums › Operations › Manufacturing › Using Mode of Data as Data Point
- This topic has 7 replies, 5 voices, and was last updated 11 years, 10 months ago by
SOS.
-
AuthorPosts
-
September 30, 2010 at 2:31 pm #53601
RappuhnParticipant@crappuhnInclude @crappuhn in your post and this person will
be notified via email.I have run into an issue that I have never come across before. I am working with a brake pad manufacturer that uses a Grindosonic tester to measure parts. They drop a “bb” onto a part 5 times and then record the reading that is most common as the data point for that part.
I have never heard of using the mode as a data point before and was wondering if someone could explain why you would use this rather than averaging the 5 data points.
0October 8, 2010 at 11:14 am #190835
SomogyiParticipant@bsomogyiInclude @bsomogyi in your post and this person will
be notified via email.Hi,
I don’t want to leave you alone with this problem, however I don’t realy understand your test method.
I see only sense of using mode if you have high risk of “false reading”. So out of 5 measurements if you have 1 or 2 improper measurements, could bias your mean but not your mode. It’s not clear for me if this measurement rule is set on your test equipment or it’s rather a procedure in your company. You may asked the tester manufacturer and/or the professional who set this test method (a Test Engineer maybe). Did they say anything?Balazs
0October 8, 2010 at 11:39 am #190837Mean is useful only when your data is normally distributed.
If data is skewed, then Median or Mode is used.Test your data first for normalcy.
Anupam
0October 8, 2010 at 10:11 pm #190838Rabos,
Give me one instance where you’ve ever seen mode used instead of mean.
You don’t know what you are talking about.
0October 9, 2010 at 4:50 am #190839Stan,
Since, most of the statistical analysis and tools are based on normal (..a theoritical..) distribution.
The data collected in real-life situation is transformed to normal distribution.
In normal distribution Mean, Mode & Median concide and therefore it does not matter which one is used as centre line or point.For Example: If you collect the data for Shoe Size, you would notice the peak occuring at Mode and not at Mean.
You would be grossly wrong if you do your analysis based on Mean.Statisticians found a round about solution for these situations i.e. Centre Limit Theorem.
Hope it clarifies
Anupam
0October 11, 2010 at 2:44 am #190847Answer the question. You don’t know what you are talking about.
0October 11, 2010 at 2:44 am #190848Answer the question. You don’t know what you are talking about.
0October 12, 2010 at 5:24 pm #190862This example may be considered if your understand, because I do not have an engineering back-ground to understand the problem fully. You shall go through the below paragraph.
Assuming the data is discrete.
A manager in a T-shirt show room will be interested only in knowing which size of T-Shirt is moving fastly. Is it Small (S), Medium (M), Large (L), XL or XXL. Based on this he will be placing the next order. There is no use, in summing or finding the average t-shirts sold. This is not going to help him in procuring new orders. Here fastly moving t-shirts represents the mode.
Whereas in a Continuous Data, the probability of obtaining a repeated value is very low. if your data is not normally distributed, mean will not be meaningful, instead Median can be considered. Here they have gone for Mode, because it is frequently repeated, actually they are supposed to represent only Median, because the concept behind the Median cannot be explained to non-quality professionals, they have gone for mode. Or they might have drawn a histogram, mode would have appeared there.
Therefore, if the data is continuous mode rarely appears, check this first.
0 -
AuthorPosts
You must be logged in to reply to this topic.