# Resource Utilzation % – Discrete or Continuous?

Six Sigma – iSixSigma Forums Old Forums General Resource Utilzation % – Discrete or Continuous?

Viewing 8 posts - 1 through 8 (of 8 total)
• Author
Posts
• #48960

mand
Member

Dear All,
We are doing a project on Resource utilization,  I have a confusion regarding the Data Type of Y  Resource Utilization % metric. As per  my understanding its a continuous data. As the metric is % based, it gives us a real number and that does not fall into any of the discrete category like Boolean or count. Also the data can be represented on a time scale (month on month basis). So the data has to be continuous.
Formula Used for calculating the Resource Utilization % = Actually Billed ELTs / Actually Billable ELTs * 100.
Example: Say 70 people joined as Resources up till July 07.
Now as per our Defect Definition: People not getting billed within 5 months of hiring are a defect.
Assume only 30 Resources are getting billed in December 07 ( Where as, as per our defect definition; all 70 Resources should start getting billed in December  as they have completed 5 months in the system); hence forth our Resource Utilization % = 30/70 * 100 = 42.85%
Like wise we have calculated the Resource Utilization % from May 07 onwards (As we have the count of Resource joined since January 07  who should get billed in May 07).
According to one of my friend (BB)  it is a Discrete Data, as we have pulled this data from the Count figures –  but still i don’t understand the reason, Why? Cos i still remeber frm the BB training class, one way of converting Discrete data to Continuous is to convert it in terms of % over time scale.
Any comments on this will be appreciated, specially if it’s a discrete data, WHY?
Regards,
Sam

0
#166370

Pai
Participant

Hi,
Technically this is descrete as its source is count.
If your counts are high then you can treat it as continuous.
See your metric jump with low count eg  if Actually Billable ELTs  is 70 then each unit change in the numerator will be 1.4 %!! Now it depends if you can live with is variation.
Regards

0
#166371

mand
Member

Hi Rajesh,
Thanks for your post, would like to have more clarity on “metric jump with low count eg  if Actually Billable ELTs  is 70 then each unit change in the numerator will be 1.4 %!! Now it depends if you can live with this variation.” can you please elaborate on this.
Thanks n Regards,
Sam

0
#166372

Pai
Participant

Sam,
I used the example in your post of 30/70=42.85%. If 30 changes to 29 the value of your metric changes to 41.428% ( a diff of 1.4%). Now analyse what will happen if you set a target value to this measure with a tolerance of 1% !! Your measuring system will have variation more than the tolerance.
Why counts are not always the best…here is more…
Say at the org level you have good counts which will kind of stabilize the metric…however when you do segmentation, the problem may crop up once again.
Suggestion could be use billable days or hours…a simple multiplication will bring some respite.
Regards,

0
#166373

Pai
Participant

The suggestion will not work !!
Regards,

0
#166376

mand
Member

Hey Rajesh,
thanks for ur wise words,
Is their any other thumb rule to identify the data type,
Especially most of the times % data is derived from the discrete data – so can we conclude that if count data is converted to % – it should always be treated as Discrete data.
n the other thing is how to go abt calculating process capability using Minitab from this discrete data set or should we go ahead with the Z calculation using DPMO?
Any inputs are always appreciated.
Regards,
Sam

0
#166381

HF Chris
Participant

Sam,A couple of things that come to mind:1. Tracking ratios of people not receiving billing after 5 months helps you fix what?2. What is delayed in the transactional process? Certification or the billing process?3. Why not track time to certification (continuous data), look for variation and then understand what is causing the variation?4. Why not track the billing process and apply a threshold (upper limit for a critical path)?This like to tracking fatalities and lost time injuries, lagging metrics are just that, too late.

0
#166430

Pai
Participant

Hi, Like I said earlier, if counts are large, then % can be considered as continuous….and you can use minitab or DPMO (be consistent in before and after)
My personal opinion, Six Sigma is be about improvements and not statistics.
Regards,

0
Viewing 8 posts - 1 through 8 (of 8 total)

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