I paid over $3000 usd for a this class in college, and it was not a good class. 18 minutes of this video, FREE, and I understand these charts better. thank you
Hi Andy, Thank you very much for your awesome efforts and contributions on helping and educating people through your SPC tutorials. Can you please explain the purpose of the UCL's and LCL's in your Np, P, C and U tutorial, as i always thought the UCL's/LCL's were determined from the customer? From your tutorial, the results will always show as "good"? Thank you.
Great question!! Okay, so your customer always determines your specifications (USL / LSL). Your specification limits are not the same as your control limits. Your control limits reflect the natural variation in your process and are calculated based on that variation. The specification limits again come from the customer and should be based on the functionality / performance of the product and are not based on the process
Are the TH-cam videos that you provide enough to pass the exam. I am definitely still considering doing your course asap, but I need some more information.
Hey There Timeng! so be honest, no, the youtube videos are not enough. I probably only cover 20% of the CQE body of knowledge for free here on TH-cam. Which is why I recommend the course, which covers every topic, and gives you every resource you need to be successful on exam day!
Great question, okay so think about how that calculation is made. We take the natural variation from the process itself, and we go out 3 standard deviations to create those upper/lower control limits. So we call those UCL/LCL the "Voice of the Process" because they reflect the natural variation in the process.
Hey There! Okay, so usually control charts should be considered an on-going process. Generally, you want to wait to collect around 25 sequential sub-groups of data before calculating control limits. Then, the frequency of which you collect another sub-group of data is really up to you.
This is by far the best explanation on Attribute Data Control Charts that I have managed to find - thank you!...But I wondered whether you could provide any further detail on exactly what we mean when we are talking about 'subgroups' / 'lots' - ideally with an example. I was initially thinking this could refer to aspects like 'factory location', or specific 'parts' (e.g parts of a car for 'defectives' example), but surely for the control chart to be plotting meaningful performance of a partcicular process over time, the data points need to be based on output of the same/comparable process?
Hi there, thank you for the video. By the way can you let me know how do i decide what is the best sample size for an attribute chart?. Which tools do i use to "calculate" or estimate the best suitable sample size for attribute chart.
Hey Albert! I would personally recommend trying to build alot of the QC tools in excel yourself - it's a great way to confirm that you understand how to perform all of the calculations, but also test your excel skills in terms of building charts. And for some of the QC tools (flow charts, C&E diagram) I would first recommend building these tools with post-it notes, and then converting them into a digital product using something like powerpoint.
Yes, thank you so much for suggestions of self learning again. I have threw all text books as small living place, the books which I bought in2006 for QM MSc in Uni of HK, some formulas I have really forgotten. Now I want to relearning for using in my job
Hi .. can we implement on process of a providing services not a physicals product and what would be the difference of measuring a service process rather than a production process ? ... Thank you in advance
Also is there an attribute chart that captures when I have more than pass/fail? I have a heat pipe soldering requirement that recieves X-ray inspection and the customers spec has an target / acceptable / fail criteria and it seems like it would valuable to capture when it is in acceptable condition in addition to target and fail.
Thanks Ardan! Yes, you can estimate process capability with discrete data - although that's probably a whole seperate video, but basically you estimate your DPMO, and most process capability tables correlate DPMO with Cpk
Hello Mr. Robertson, I have a question following your very clear and helpful videos. Once I calculated CL, UCL and LCL the data belong to the "past". Now a new lot was produced. How do I incorporate the new data in the CL's? After how many new lots should I update the limits? Do I disregard the first lots from the calculations? What is the periodic update recommended ?
Hey Jose, great question, in fact I would argue that this is one of the trickiest parts of SPC, because there isn't a specific answer to that question. It really depends a lot on your process. Technically, if your process is "stable" and "in control" you shouldn't expect much change in your range chart, (or the width of your X-bar chart). This is especially true if you setup your sub-groups correctly to be "rationale" in that they only capture common cause variation. But what can change is the centerline of your process. You might have long term sources of common cause variation that might change the center of your process, and force a recalculation of the limits. Like I mentioned though, there aren't any hard and fast rules around this and it depends on your process.
@@CQEAcademy Thank you Andy. It seems everything actually (past, present and future) depends on the process performance. Considering that special cause cases the updade should be as soon as detection occurs, I am still looking for a reasonable or recommended "moving update number" of samples on common cause processes. Is there any?
@@josechvaicer7328 Not that I'm familiar with Jose. What I would recommend is an experiment. Choose a set number of sub-groups and re-calculate the control limits after that set number. If you find that the control limits don't change often, increase that number. If they change more often (and you still feel confident that the process is only experiencing common cause variation) then decrease that number.
Hii. Your video is really great and extremely helpfull. I just wanted to ask you if you could tell me if there is any possibility to find real datasests from real companies to use for a calculation?
You videos are always so clear and easy to understand - thank you so much for having created these videos and helping me study and understand various concepts for my lean six sigma green belt exam!!
@@CQEAcademy So cool!!! I played basketball at Adams State down in Alamosa, so it stood out to me. A couple friends of mine are CSU alum from 2014/2015 graduating classes. Thank you for all your knowledge and insights. I feel even more of a personal connection! All the best! Cheers! 🥂
Hi! I work in Healthcare Facility and we count the number of patients visit emergency department on monthly basis. Which type of control chart recommended to be used and what category these data is classified
Hey Maher, if you're just counting the number of patients visiting, this would be equivalent to counting defect items - 1 person is a single visitor. And since you're not really sampling, you're simply just counting, i'd use an NP chart. If you were trending the number of "ailments" per month, where one visitor could have multiple ailments, that would be like counting defects. -Andy
Thank you for your explanation in the video. Can i used some prinsip if i have the data of finished good, (non defect product)? Or maybe is there any other way to check control limit of those data? Thank you
Hello , I really enjoy your videos, say you have 80 observations, each observe a number of people waiting in a line, which Control Chart would you use for it? if I try np I won't be able to get the P-bar
Hey Faisal, will you have the same 80 people in line every day? If not, and the daily sample size changes, you'll want to use either the U chart or the P chart.
Hi, thank you for making these videos! Your videos are truly easily to understand and super well structured. One question: you mention that the NP & P Chart are based on binomial distributions and that C / U Chart are based on Poisson distribution. Why is that the case? It would be terrific to get any pointers here. Thank you again!
Hey Andreas!! Great question, so the NP and P chart are based on the binomial distribution, because one of the assumptions is that there are a fixed number of outcomes to any experiment - and for control charts that means conforming or non-conforming. For a C / U chart, there can be multiple defects, which means that this assumption in the binomial distribution cannot be met, so we must move to the poisson distribution where multiple outcomes are possible (multiple defects per parts being inspected).
@@CQEAcademyboth P and u chart the LCL and UCL am getting different values, what’s the value to be considered as u-bar or P-bar, for example in the ucl of u bar example, if I recheck u-bar is 750.69 in that equation if I reverse calculate
Hey Dhandapani, great question! Okay, so think about what the Y-Axis is for the NP chart. We're counting the number of defective units, and it is impossible to have a negative number of defects. If the calculation for the LCL for an NP chart is a negative number, you can simply drop the LCL or set it at zero. Depending on how close the LCL is to zero, having zero defective units might be cause for alarm (under-inspection) or it might just be simple, random variation. -Andy
Regarding the defect vs defectives… I thought that the unit wouldn’t be considered defective unless the defects prevent the product from functioning. Meaning just because a unit has a defect on it, it wouldn’t be considered defective.
Hey There! Okay so that definition I provided is the generic definition of a defective item. It's possible that your organization might have a slightly different definition that's risk based and considers the impact to functionality. If your organization has a rule that allows you exclude a failure mode from being "defective" based on its impact to the customer, you could simply exclude that from consideration while executing your control chart. The real key part of that definition is whether or not the entire unit is counted as good or bad - which allows you to use the Binomial distribution and the P-chart or the NP Chart.
Hey, I was just wondering why 3 in every UCL and LCL formula? Is there an specific reason for it to be 3 times the square root of the parameters? Can the 3 be changed in certain situations? Thx in advance!
Hey David, the 3 ensures that we capture the +/- 3 sigma. That term to the right of the 3 reflects the standard deviation associated with those control charts, so when we multiple by 3, we get +/- 3-sigma, which is the expected, normal variation. It's not very standard to change the 3. If you were to reduce it, you'd increase the "alpha risk" associated with your control charts, meaning that you'd be getting "out of control points" when the process was really just experiencing normal cause variation.
Excellent video! Now, since SPC charts always look to the past (what HAS happen) what would be the guidelines for establishing the "looking ahead" period or days to renew the graphs?
![PASULOL รามเกียรติ์ ตอนที่ 4 นนทกพบรัก [Ramakien Ep.4: Nonthok in love]](http://i.ytimg.com/vi/tMUQgmZzf_A/mqdefault.jpg)
I paid over $3000 usd for a this class in college, and it was not a good class. 18 minutes of this video, FREE, and I understand these charts better. thank you
This vid is so underrated
Wow, thanks so much Satya!
Perfect
Thank you!
Hi Andy,
Thank you very much for your awesome efforts and contributions on helping and educating people through your SPC tutorials.
Can you please explain the purpose of the UCL's and LCL's in your Np, P, C and U tutorial, as i always thought the UCL's/LCL's were determined from the customer?
From your tutorial, the results will always show as "good"?
Thank you.
Great question!!
Okay, so your customer always determines your specifications (USL / LSL).
Your specification limits are not the same as your control limits.
Your control limits reflect the natural variation in your process and are calculated based on that variation.
The specification limits again come from the customer and should be based on the functionality / performance of the product and are not based on the process
Thankyou
Are the TH-cam videos that you provide enough to pass the exam. I am definitely still considering doing your course asap, but I need some more information.
Hey There Timeng!
so be honest, no, the youtube videos are not enough. I probably only cover 20% of the CQE body of knowledge for free here on TH-cam.
Which is why I recommend the course, which covers every topic, and gives you every resource you need to be successful on exam day!
This is the BEST video in internet to understand Attribute Charts with Examples. I wish i found this video first. Thanks
hey thanks for the video. where could I find the cheat sheet?
You're welcome! You can find that resource here:
cqeacademy.com/FreeCheatSheet
great video.
something i don't understand. don't i determine what USL and LSL are? why it determined by the results of the process?
Great question, okay so think about how that calculation is made. We take the natural variation from the process itself, and we go out 3 standard deviations to create those upper/lower control limits.
So we call those UCL/LCL the "Voice of the Process" because they reflect the natural variation in the process.
Nice presentation. I was very helpful.
Thanks!
Great explaination
Thanks!!!
I've been struggling to understand the charts and I finally found the right video. Thank you
Great. You made this so simple. 👌
Thank you for detail information. How to determine process monitoring and control chart period ?
Hey There! Okay, so usually control charts should be considered an on-going process. Generally, you want to wait to collect around 25 sequential sub-groups of data before calculating control limits.
Then, the frequency of which you collect another sub-group of data is really up to you.
Million likes for your videos. Really appreciated.
Thanks Nguyen!
This video was REALLY helpful and detailed, thanks
Glad it was helpful!
This video is extremely useful.. It helped us with our tests.. thanks a ton!!
You're absolutely welcome, I'm glad the videos were so helpful!
@@CQEAcademy honored by your reply.. could you please upload videos on two tailed F-tests or ANOVA please? With examples
..
@@user-yq9kr9sy9i Hey there K, I'll add an ANOVA video to the to-do list
Hi, can i know what is the t chart in SPC? Is there any others name for it and what is it? Look forward for ur reply, thank you.
Best Explanation, great job!!. Thank you!!
This is by far the best explanation on Attribute Data Control Charts that I have managed to find - thank you!...But I wondered whether you could provide any further detail on exactly what we mean when we are talking about 'subgroups' / 'lots' - ideally with an example. I was initially thinking this could refer to aspects like 'factory location', or specific 'parts' (e.g parts of a car for 'defectives' example), but surely for the control chart to be plotting meaningful performance of a partcicular process over time, the data points need to be based on output of the same/comparable process?
thank you for this video
you're welcome Jitendra!
Very well done explanation of attribute control charts..
Thanks Victor!!! i'm glad you liked it!
-Andy
Amazing content. Thanks for the video.
You really really help me thank you very much I have exam tomorrow 🤩🙏
You're welcome!!! I'm happy to help!
Excellent explanation.Thank you very much
You're welcome Prabhakaran!!!
-Andy
I'm just wondering how can u-chart take variable sample size?
Hi there, thank you for the video. By the way can you let me know how do i decide what is the best sample size for an attribute chart?. Which tools do i use to "calculate" or estimate the best suitable sample size for attribute chart.
Hi Andy, on your discussion for u-chart, your UCL/LCL has n-bar but in other books I see its n(i), not n-bar. Could you clarify?
That's great
You're welcome Jannatul!!!
Sum of subgroup quantities with an s!
Sir please explain scatter diagram from basic ,like which data is used for scatter diagram .how the 2 variables are correlated with each other ?
That's a good suggestion, I can add that to my to-do list.
Thank you Sir! Another great video
You're welcome!!!!!!
Can I get the templates of 7 old QC charts with formulas from internet that I may input data to seeing the results.
Hey Albert! I would personally recommend trying to build alot of the QC tools in excel yourself - it's a great way to confirm that you understand how to perform all of the calculations, but also test your excel skills in terms of building charts.
And for some of the QC tools (flow charts, C&E diagram) I would first recommend building these tools with post-it notes, and then converting them into a digital product using something like powerpoint.
Yes, thank you so much for suggestions of self learning again. I have threw all text books as small living place, the books which I bought in2006 for QM MSc in Uni of HK, some formulas I have really forgotten. Now I want to relearning for using in my job
Hi .. can we implement on process of a providing services not a physicals product and what would be the difference of measuring a service process rather than a production process ? ... Thank you in advance
13:36 does that also apply to the other Attributes Data Control Chart?
Yes
What if your lot sizes change but your samples stay the same? Should that be treated the same as your sample sizes varying?
Also is there an attribute chart that captures when I have more than pass/fail? I have a heat pipe soldering requirement that recieves X-ray inspection and the customers spec has an target / acceptable / fail criteria and it seems like it would valuable to capture when it is in acceptable condition in addition to target and fail.
Hey Mark, yes I would use the p chart or the u chart in this scenario
Wow easy to understand,
Hey Sir, Could you calculate Process Capability for Attribute Data? (For C chart example)
Thanks Ardan! Yes, you can estimate process capability with discrete data - although that's probably a whole seperate video, but basically you estimate your DPMO, and most process capability tables correlate DPMO with Cpk
Hello Mr. Robertson, I have a question following your very clear and helpful videos. Once I calculated CL, UCL and LCL the data belong to the "past". Now a new lot was produced. How do I incorporate the new data in the CL's? After how many new lots should I update the limits? Do I disregard the first lots from the calculations? What is the periodic update recommended ?
Hey Jose, great question, in fact I would argue that this is one of the trickiest parts of SPC, because there isn't a specific answer to that question. It really depends a lot on your process.
Technically, if your process is "stable" and "in control" you shouldn't expect much change in your range chart, (or the width of your X-bar chart). This is especially true if you setup your sub-groups correctly to be "rationale" in that they only capture common cause variation. But what can change is the centerline of your process. You might have long term sources of common cause variation that might change the center of your process, and force a recalculation of the limits. Like I mentioned though, there aren't any hard and fast rules around this and it depends on your process.
@@CQEAcademy Thank you Andy. It seems everything actually (past, present and future) depends on the process performance. Considering that special cause cases the updade should be as soon as detection occurs, I am still looking for a reasonable or recommended "moving update number" of samples on common cause processes. Is there any?
@@josechvaicer7328 Not that I'm familiar with Jose. What I would recommend is an experiment. Choose a set number of sub-groups and re-calculate the control limits after that set number. If you find that the control limits don't change often, increase that number. If they change more often (and you still feel confident that the process is only experiencing common cause variation) then decrease that number.
@@CQEAcademy Trial and error.....should give the answer. Thank you for the practical hint.
Hii. Your video is really great and extremely helpfull. I just wanted to ask you if you could tell me if there is any possibility to find real datasests from real companies to use for a calculation?
Hey Mila!
Unfortunately most companies don't publish or share their SPC data, so I'm not aware of any real datasets to analyze
It was a really great explanation Thank You
I'm glad you liked it Siva!!!
-Andy
You videos are always so clear and easy to understand - thank you so much for having created these videos and helping me study and understand various concepts for my lean six sigma green belt exam!!
You're welcome, and I appreciate the great feedback!
Best of luck on your green belt exam!
is that a colorado state university rams flag in the background??
Good eye Davon!!! I am a proud CSU Ram!!!
@@CQEAcademy So cool!!! I played basketball at Adams State down in Alamosa, so it stood out to me. A couple friends of mine are CSU alum from 2014/2015 graduating classes. Thank you for all your knowledge and insights. I feel even more of a personal connection! All the best! Cheers! 🥂
@@davongrant1368 Same to you Davon!!! And I appreciate the comment and the great feedback!!!
Hi! I work in Healthcare Facility and we count the number of patients visit emergency department on monthly basis. Which type of control chart recommended to be used and what category these data is classified
Hey Maher, if you're just counting the number of patients visiting, this would be equivalent to counting defect items - 1 person is a single visitor.
And since you're not really sampling, you're simply just counting, i'd use an NP chart.
If you were trending the number of "ailments" per month, where one visitor could have multiple ailments, that would be like counting defects.
-Andy
Would love a video on variable control chart types and how to choose between them! Love your videos
Check out this video: th-cam.com/video/Aj7lJLR-7b4/w-d-xo.html
These videos are the best on TH-cam. Thanks for the quality content and for helping me pass the green belt exam!
Wow thanks so much Sara!!! You're welcome!!
By the way, for anyone else out there preparing for the green belt exam - check out greenbeltacademy.com!
I like all your content Andy, it's really helpful
Thanks so much!!!
One of the best out of few I watched. Excellent explanation. Nicely done.
a consize method for remembering
Thanks!!!
Thank you for your explanation in the video. Can i used some prinsip if i have the data of finished good, (non defect product)? Or maybe is there any other way to check control limit of those data? Thank you
Hello , I really enjoy your videos, say you have 80 observations, each observe a number of people waiting in a line, which Control Chart would you use for it? if I try np I won't be able to get the P-bar
Hey Faisal, will you have the same 80 people in line every day?
If not, and the daily sample size changes, you'll want to use either the U chart or the P chart.
How can you help me create a control chart for Pressure injury
Hi, thank you for making these videos! Your videos are truly easily to understand and super well structured.
One question: you mention that the NP & P Chart are based on binomial distributions and that C / U Chart are based on Poisson distribution. Why is that the case? It would be terrific to get any pointers here.
Thank you again!
Hey Andreas!! Great question, so the NP and P chart are based on the binomial distribution, because one of the assumptions is that there are a fixed number of outcomes to any experiment - and for control charts that means conforming or non-conforming.
For a C / U chart, there can be multiple defects, which means that this assumption in the binomial distribution cannot be met, so we must move to the poisson distribution where multiple outcomes are possible (multiple defects per parts being inspected).
really appreciate your efforts! thanks so much!
You're welcome!!!
Butt🥺 why when I calculate I'm getting math error 😢
Hmmm that's odd, at which point are you getting an error???
@@CQEAcademyboth P and u chart the LCL and UCL am getting different values, what’s the value to be considered as u-bar or P-bar, for example in the ucl of u bar example, if I recheck u-bar is 750.69 in that equation if I reverse calculate
Hey @@raghudotnet U-bar should be 7.5% and is calculated as 117 defects, divided by 1,560 units inspected.
Can Lower control limit could be negative in NP chart ??
Hey Dhandapani, great question!
Okay, so think about what the Y-Axis is for the NP chart.
We're counting the number of defective units, and it is impossible to have a negative number of defects.
If the calculation for the LCL for an NP chart is a negative number, you can simply drop the LCL or set it at zero.
Depending on how close the LCL is to zero, having zero defective units might be cause for alarm (under-inspection) or it might just be simple, random variation.
-Andy
Regarding the defect vs defectives… I thought that the unit wouldn’t be considered defective unless the defects prevent the product from functioning. Meaning just because a unit has a defect on it, it wouldn’t be considered defective.
Hey There!
Okay so that definition I provided is the generic definition of a defective item.
It's possible that your organization might have a slightly different definition that's risk based and considers the impact to functionality.
If your organization has a rule that allows you exclude a failure mode from being "defective" based on its impact to the customer, you could simply exclude that from consideration while executing your control chart.
The real key part of that definition is whether or not the entire unit is counted as good or bad - which allows you to use the Binomial distribution and the P-chart or the NP Chart.
@@andyrobertson566 thanks for the quick response!
Hey, I was just wondering why 3 in every UCL and LCL formula? Is there an specific reason for it to be 3 times the square root of the parameters? Can the 3 be changed in certain situations? Thx in advance!
Hey David, the 3 ensures that we capture the +/- 3 sigma.
That term to the right of the 3 reflects the standard deviation associated with those control charts, so when we multiple by 3, we get +/- 3-sigma, which is the expected, normal variation.
It's not very standard to change the 3. If you were to reduce it, you'd increase the "alpha risk" associated with your control charts, meaning that you'd be getting "out of control points" when the process was really just experiencing normal cause variation.
Hi! Where can I get a powerpoint presentation just like this one? Thank you!
Hey Ayessa!!! I made this presentation myself, it's sort of my secret sauce :)
C Chart UCL and LCC values are wrong
Hey Jaya, what values do you calculate?
Excellent video! Now, since SPC charts always look to the past (what HAS happen) what would be the guidelines for establishing the "looking ahead" period or days to renew the graphs?