# Time – Discrete or continous

Six Sigma – iSixSigma › Forums › Old Forums › General › Time – Discrete or continous

- This topic has 21 replies, 13 voices, and was last updated 14 years, 5 months ago by BuckDaddyToeheadedDimwit.

- AuthorPosts
- August 30, 2005 at 6:05 am #40523
quick question is time discrete or continous data ?

0August 30, 2005 at 8:27 am #126036depends on what’s the unit of measure you are using. i would consider time, if measured in seconds or hours, as a continuous data for all practical purposes.

Just my opinion.0August 30, 2005 at 9:26 am #126038discrete, one can not say 1.2 time, 1.345 time etc. Just my opinion

0August 30, 2005 at 10:00 am #126041Time is a metric not a unit of measure. You would never measure time as 1.2 time or 1.345 time. It would be 10.2 mins or 1136 secs etc.. It is a continuous data.

0August 30, 2005 at 2:19 pm #126052Austin:Time is and always has been continuous data, also called variable data. People often make a mistake and turn it into discrete data by classifying events into bins, eg. ‘late,’ ‘on-time,’ and ‘early.’For example, aircraft arrival times for flights in and out of Chicago (ORD) are classified is such a manner. What exactly does ‘63% on-time’ mean?BTDT

0August 30, 2005 at 3:28 pm #126057

Anton JavierParticipant@Anton-Javier**Include @Anton-Javier in your post and this person will**

be notified via email.Austin, there is what you call as the “half test”.

If the metric is something which you can split, like half an hour, half a minute, or half pound, then it is adjudged as continuous or variable.

On the other hand, if you cannot split it, like half accept, or half good, then it fails the “Half Test”. You now can conclude that it is an attribute data.

I hope this helps.

Anton

0August 30, 2005 at 3:51 pm #126060Time is a function of how it is used. For instance is it Seconds, Minutes, Hours, Days or Years; or is it Half a Day, or Half a Month, Or Half a Year. If you use such terms as Half a Day then you have to define the “Half”; is it 4 hours, (In Reference to shift) or 12hrs(In Reference to the whole Day) at any rate it is Continuous Data.

CT0August 30, 2005 at 3:54 pm #12606163% on time in and out of O’Hare means it was a good day, regardless of data classification/categorization. Sometimes even inappropriate manipulation won’t help your numbers.

Vinny

0August 30, 2005 at 4:15 pm #126062Vinny:For ORD, this is the baseline.BTW – they don’t lose your luggage, they only ‘misdirect’ it.Cheers, BTDT (7 days with only carry-on)

0August 30, 2005 at 6:07 pm #126074Time is discrete if you having a lot of fun and continuous if you are lonely and all by yourself

0August 30, 2005 at 6:12 pm #126076

The One BillMember@The-One-Bill**Include @The-One-Bill in your post and this person will**

be notified via email.Vin-

And, 20% of the time you providing anything to this forum of value would be amazing!0August 30, 2005 at 6:15 pm #126077

The One BillMember@The-One-Bill**Include @The-One-Bill in your post and this person will**

be notified via email.What if you had 25 individual months of data to evaluate the sales trend of a business? Continuous or discrete?

Don’t have an opinion0August 30, 2005 at 6:25 pm #126079What are you measuring? I assume you have 25 months of sales data. In that case, sales would likely be continuous if you are talking about dollars but possibly discrete if you are counting number of sales. The fact that you have 25 months does not relate to the previous issue of whether time was continuous or discrete.

0August 30, 2005 at 6:34 pm #126080

The One BillMember@The-One-Bill**Include @The-One-Bill in your post and this person will**

be notified via email.Explain to me how the period the data were collected within has any bearing on whether it’s continuous or discrete…

What if I told you these data were collected in 25 hours. Would that change anything? What is the real qualification for interval data? You’re the guru.. I’m just the grasshopper!

0August 30, 2005 at 7:27 pm #126083The One Bill:It would be best to have 25 months of the transactions for more extensive analysis, if your interest was something like time between transactions for capacity planning.If you have 25 months of consolidated monthly data, then you have 25 measurements of dollars/margin. The Y data (dollars) is still continuous, while the time data are sufficient for doing regression to identify your trend.You may even have enough data to fit a seasonal+trend model to the data.BDTD

0August 30, 2005 at 7:33 pm #126085Time is Variable Data or Quantitative meaning continuous (Decimal Subdivisions are Meaningful)-any measure of time in Seconds, Minutes, Hours, ECT.

Time can Also be Attribute Data or Qualitative when it can be counted. Example, Number of Days to perform an order transaction.

IF the number of days to perform a transaction is broken down into measurable time then it becomes Continuous. Days on the other hand is a general count of 24hr periods.

The Key to Descrete or Continuous is Counting Data or Measuring Data.

CT0August 30, 2005 at 8:13 pm #126089Great response BDTD! Too bad Darth didn’t get it!

0August 30, 2005 at 8:28 pm #126093What are you measuring? I assume you have 25 months of sales data. In that case, sales would likely be continuous if you are talking about dollars but possibly discrete if you are counting number of sales. The fact that you have 25 months does not relate to the previous issue of whether time was continuous or discrete.

Please provide insight as to what you find incorrect about this response.0August 30, 2005 at 8:37 pm #126095BTDT,

Since somebody likes your answer better than mine maybe you can expand upon your suggestion that you can do a regression with sales (Y) as continuous and the months as the X. Unless you are suggesting use of a logistic regression are you not then assuming that the months are continuous? Should the poster not do a control chart first? That would allow him/her to get a quick snapshot of the process and an ability to see a trend.0August 30, 2005 at 9:50 pm #126104The type of data does not change with the size of your sample or the length of time taken to collect it. It is simply the format or medium you chose to work in. Think of continuous data as letters and attribute data as words. The former is much more useful in communicating more with less.

Although, I believe you can approach continuous-like characterisitics if your attribute data contains adequate discrimmination (e.g. The popular “temp gauge” on surveys that ask you how you feel about something on 1-100 scale). This is arguable at best, if not all together wrong. Stand by for a more informed, if not downright hostile rebuttle.

As a newbie, one option would be to always choose continuous data as your data type when crafting your collection and sampling plan….GE essentially does this with their GB and BB…using this type of “if-then” checklist mentality early on can be helpful as you gain experience in the skill set. Choose robust, universal tools that work in one environment, and then work only in that environment for a time. Good luck.0August 30, 2005 at 10:29 pm #126107DrD:A complete explanation of the true nature of the dataSET[sic] would help everyone considering a response to this post. Most of us have assumed that the time data is X, and the Y data is something else like dollars.A Run chart is always the best, first thing to do with a set of data. In this case, we would be doing a run chart on dollars. If I assumed that the data was monthly consolidated data, then a regression could be of dollars(Y) versus month(X). What disturbs me about the typical time-series data analysis is that it assumes that the time series data is from fixed intervals of time, but that makes the time series a bit easier to analyze. We can stay away from Fourier analysis.The assumption of ordinary least squares is that the error term is restricted to errors in Y only. The fact that the month data (X) corresponds to fixed months (1,2,3, etc.) is not a problem; there is no reason why new data could not be collected every two weeks and still be incorporated into ongoing analysis.Logistic regression would only be required if the responding variable (Y) was discrete. In our example above, the time data is X. I can not think of an example right now where the responding variable would be time AND fixed values such as months.Whether monthly data is fixed, discrete, or merely coarsely sampled, continuous data shows up in the calculation of such cases as monthly salary and mortgage payments in spite of different months having different numbers of days. Even a single bank will use different definitions depending on the situation. They will calculate and charge mortgage interest as if each month is the same length, while your savings account has interest calculated at the end of each day.I have no idea why someone might like my answer more than yours. It must be my karma, boyish charm, and razor wit.*BTDT*or the iSixSigma sport of consultant baiting.

0August 31, 2005 at 12:51 am #126108

BuckDaddyToeheadedDimwitParticipant@BuckDaddyToeheadedDimwit**Include @BuckDaddyToeheadedDimwit in your post and this person will**

be notified via email.Sorry Darth.

He posted a comment in the wrong place. He was refering to a side conversation we had about our feminine sides, and how difficult it is to deal with them.

BuckDaddyToeheadedDimwit (BDTD)0 - AuthorPosts

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