Thank you for the video, but I wanted to point out that this is NOT the correct way to calculate control limits for control charts. In control charting the sigma symbol is not the same as typical standard deviation. It is calculated using different methods (depending on the type of the chart) that takes into consideration the relative shift of the process over time. This is by design and it is very critical that sigma is calculated correctly for a control chart to be useful. In your case, sigma should be calculated as [sigma = MR/d2] .....where MR is the moving range of each successive data point and d2=1.128 (the scalar constant for control charts where sample size is n=1). I encourage people to compare both ways with real data and you will see how much different the values are. Using global standard deviation (as you have) will severely inflate the control limits and can give you the false impression that a process is in control when it is likely not. The reason 3*SD seems too far away is because it’s not the right calculation. Calculating the sigma value correctly will show many of the data points are out of control even with 3*sigma. Apologies for the rant but i see this error too often. Too many MBAs who took a stat class thinking this is what statistical process control is.
I believe for an overview this technique is alright. To see if anything went beyond the 3 standard deviations. But we would love it if you could make a video on calculating it the right way.
Mate, you may very well have saved my assessment from going into the fire, Thank you for your hard work at making such a easy to follow video, it is a mark of a true teacher to make the seemingly complex, simple to follow, Thank you again.
I'm struggling to get my head round one thing... why would you want your control limits to be dynamic? Imagine the scenario where you have a data point that should sit just outside your CL. But because the CL is dynamic, it shifts so the new data point is now within the CL. It's kind of a self fulfilling prophecy! Shouldn't you use historical data to establish fixed control limits, then see where new data points fall in relation to those limits? Thanks in advance 🙂
Thank you for the video! Helped me so much. Can you get to the same result without the inputs in chart form? What if I have the dates and values across a horizontal row?
I have 480 time measurements per day for 30 days, if I calculate the average per day and out of this mean, find the standard deviation it would be correct? Should I use x- chart and s-chart or r-chart?