1
$\begingroup$

I have a column chart that looks like this: enter image description here
The chart is not well labeled because of data privacy and security issues. However, I can give you some information about the data behind this chart.

Let's consider a scenario where there are machines producing something. Sometimes events occur when a machine is in a production process. Here, 'events' are bad occurrences that stop a machine from doing its work.

The horizontal axis represents configurations or settings (all configurations concatenated into a string based code) on a machine, and the vertical axis represents the rate at which events occurred while using a specific configuration.

The rate of event occurrence (event_rate) is calculated as the number of events occurred while using a specific configuration divided by the sum of all productivity when that configuration was used.

The distribution of Event Rates is as depicted in the below chart: enter image description here My Question:
I understand that this distribution is heavily right skewed.

  1. Apart from deducing that most configurations do not have a very high event_rate (which is good news), is there any other interpretation that I can make from this chart?
  2. Other than heavily right skewed, is there any other name for this type of distribution? In other words, does this distribution classify as 'exponential' or 'log-normal' or others? Please provide explanation why.
  3. Any other insight that you can provide by looking at this chart is also appreciated.

Note: This question has been edited to add the Event Rate distribution chart and change the terminology accordingly

$\endgroup$
  • 3
    $\begingroup$ This is not a distribution: it's merely a bar chart. (By definition, a distribution assigns probabilities or frequencies to intervals of numbers or more generally to higher-dimensional analogs of such intervals.) As such, it doesn't tell us much of anything at all, only that some configurations have higher frequencies than others. $\endgroup$ – whuber Jul 8 '16 at 21:24
  • $\begingroup$ @whuber Sorry for my wrong usage of the word distribution. I realize that it is not a distribution, but only a bar chart. But, can I not still call this bar chart 'left skewed'? $\endgroup$ – Minu Jul 8 '16 at 21:33
  • 2
    $\begingroup$ Most likely not, unless there is some inherent natural order in the configurations as well as a sense of "size" or "width" of each configuration. Therein is the crux of the matter: from the sounds of it, a "configuration" is a nominal variable; it is not inherently quantified. $\endgroup$ – whuber Jul 8 '16 at 21:46
  • $\begingroup$ Thanks for your help on this. If I edit my question with the histogram of Event Rates and frequencies, will that help in finding more insights? $\endgroup$ – Minu Jul 8 '16 at 21:59
  • 1
    $\begingroup$ Event rates are ordered ascendingly in the distribution plot. Event rates are ordered ascendingly even in the column chart. $\endgroup$ – Minu Jul 9 '16 at 23:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.