Vishwadeep Khatri

How to do it in Excel (works for non-uniform bin widths)
Assume your sheet looks like the screenshot:
A: Frequency_Hz
B: PSD_X (g²/Hz)
C:_PSD_Y (g²/Hz)
D:_PSD_Z (g²/Hz)
Add helper columns:
E (Δf): bin width
In E3: =A3-A2 → copy down
F (area_X): trapezoid area for X
In F3: =0.5*(B3+B2)*E3 → copy down
G (area_Y): trapezoid area for Y
In G3: =0.5*(C3+C2)*E3 → copy down
H (area_Z): trapezoid area for Z
In H3: =0.5*(D3+D2)*E3 → copy down
At the bottom (replace LASTROW with your last data row):
X-axis grms:
=SQRT(SUM(F3:F_LASTROW))
Y-axis grms:
=SQRT(SUM(G3:G_LASTROW))
Z-axis grms:
=SQRT(SUM(H3:H_LASTROW))
Overall tri-axial GRMS: (vector RMS)
=SQRT( X_grms^2 + Y_grms^2 + Z_grms^2 )
Round to one decimal place (if you want a single “one-digit” figure):
=ROUND( Overall_GRMS , 1 )
Notes 
Your PSD columns must be linear g²/Hz, not dB. If they’re in dB:
convert first with PSD_linear = 10^(PSD_dB/10).
If your analyzer exported ASD (g/√Hz) instead of PSD, square it first to get g²/Hz, then use the same steps.
Only bins inside your band of interest should be included in the sum.
GRMS is in g. If you need m/s² RMS, multiply by 9.80665.
That’s it—drop those formulas in and you’ll get GRMS per axis and overall.
     

Excellent question, Senthilnathan. 

If the process is consistently generating high percentage of defects, it may pass the stability test and is considered to be in statistical control. However, it might fail badly on the Capability Assessment if customer is unhappy with the defect level. Now, I have used the word "might" because the unhappiness of customer depends on the competitive scenario or the novelty of the solution. 

To provide you an example, there is an app called "Skyscanner" for which you can compare prices for the same flights over different platforms. It is an aggregator of aggregators. In US, customers of Skyscanner are delighted because they are saving in most of their flights or stay although, the tool is not perfect and does show erroneous results sometimes. 

If Skyscanner is having a defect rate which is unusually high (compared to other aggregator apps) but still makes customer happy due to lack of competition as an aggregator of aggregators, its process may be considered stable as well as capable. 

By the way, Skyscanner works in India too and is worth a try ")

Cluster sampling is useful when each cluster is a mini-representation of the population. The idea in cluster sampling is not to take samples from each cluster but to consider an entire cluster (or a few clusters) as sample for the entire population. This means 100% data from selected cluster(s) becomes the sample and we do not need to take any smaller proportions from the selected cluster. 

Hi Dheeraj,

Your situation is indeed a common challenge many Six Sigma professionals face when there's an organizational division of professionals across methodologies. While specialization can help focus efforts, it can sometimes be limiting.
Here are some strategies to adapt:
 
Integrate Six Sigma with Automation: Identify projects where automation is the goal but needs a structured problem-solving approach. Introduce Six Sigma methods to analyze and validate the need for automation, thereby creating a justification for the project within the Six Sigma framework. Influence Change Through Data: Use the power of Six Sigma data analysis to demonstrate how a blended approach can yield superior results. If you have evidence, it may be easier to gain buy-in from leadership. Cross-functional collaboration: Engage with the other specialized teams. Offer to provide Six Sigma data analysis or process mapping that could benefit their projects, establishing a two-way street of value-add. Educate and Advocate: Keep an open dialogue with decision-makers to help them understand the merits of a more integrated approach. You could even offer to lead a pilot project combining methodologies.
Internal Community of Practice: Form or engage in a community where like-minded professionals can share best practices. This can be a great platform to introduce and discuss blended methodologies.
Escalate Sensibly: If all else fails and the restriction proves to be a genuine roadblock to meaningful improvement, it might be worth taking your concerns up the chain of command, equipped with case studies and data to make your point.
Remember, the ultimate goal is value creation, regardless of the methods used. Adaptability and working with other disciplines are crucial for any Continuous Improvement professional.

There is no correct answer provided for this question among the answers below. 

This question has two parts, and most respondents address the first reasonably. For the second part (what kind of statistical analysis can/ cannot be performed with transformed data), the correct response is as follows -
Process Stability and Capability Analysis can be carried out using transformed data. The control limits are also based on transformed data when using control charts. When using capability analysis, the specification limits are transformed too.  Hypothesis testing is not done with transformed data but only with the original data. There are enough hypothesis tests (non-parametric tests) designed for non-normal data. Imagine comparing averages of two data sets transformed through Johnson's transformation. This would be meaningless because the two data sets would have been transformed with different equations (Johnson's transformation generates an equation to transform the data). 

Excellent question, Amith.

Let me put it this way - A Quality Assurance professional who has Lean Six Sigma competence creates opportunities for growth outside the Quality Assurance domain. 
 
Many of the Lean Six Sigma tools were initiated in the Quality domain but a true and successful LSS professional applies the learning for generating better business results than ever before. 

If you see the evolution of Lean Six Sigma in the image below, you will notice how things have progressed in this domain over the years. Hope this helps. Do ask a follow up question if you want to understand more. 
 

RPA is expected to grow faster with AI getting stronger. 
 
Image Analysis and Text Analysis, for example are two areas within RPA that will be more successful as it is infused with Machine Learning capabilities. Reliable Artificial Intelligence can make more processes RPA-ready in future. Large number of processes today are considered unsuitable for RPA because of good percentage of exceptions in them. If Artificial Intelligence can reduce these exceptions significantly, or ideally eliminate them, RPA will become the preferred mechanism for such processes. 

The answers this time (given below) are good for the part 1 of the question but there are no attempts for part 2. Some efficiency metrics can be considered better than others if they have less harmful side effects. 
 
Productivity in manufacturing as number of units per month is a commonly used metric but promotes overproduction. Theory of Constraints promotes the use of a metric called Throughput which is calculated as Throughput = Sales - TVC (Totally Variable Costs). This is an efficiency metric which ensures that productivity will be considered valid only if Sales happen. And as evident, lower the variable costs, better will be the throughput. This metric removes some of the harmful side effects of a pure production volume metric. 
 
As no one has provided response to the second part,  there is no winner for this question

Whether machines can behave like humans or not has been an intriguing question for a long time. Why do we want to know this? 
 
The quest can be divided into into two parts - 
 
1. Can machines behave as rationally as humans can? Since IBM’s Deep Blue had beaten Kasparov in a game of chess, more powerful computers have come in. The latest chess software called Komodo has an Elo rating as high as 3304 which is 450 above any human. Such evaluation and mastery is certainly of interest to the excellence community. 
 
2. Can machines imitate humans and be similar even in irrational behaviours? This question is of interest to security experts who want to know if certain actions are being taken by computer (viruses, for example) instead of humans. Many times computers fail to behave irrationally and give away their presence due to patterns. As an example, computer virus can bring in thousands of fictitious members into a forum or online community by creating fake email ids and even clicking on verification emails by logging in to those email addresses. Google and other online companies have been creating anti spam solutions by figuring out patterns that the computers fail to match with humans. Most of the times computers tend to be more rational than humans as they have an algorithm driving them. Only time will tell if the fight with viruses will be won by well meaning humans or not. 

Some of tools that allow for customisation and enhancements in themselves are as follows - 
 
Kanban board - This provides endless opportunities for improvement in terms of breaking up of tasks, marking expedite lanes, adding tasks on kanban cards etc. Online Kanban board platforms provide even more methods of customisation and tool improvement. 
 
Future VSM - The extent and format of details that are to be captured in a future VSM have unlimited opportunities for improvement. 
 
Andon Boards - Andon boards have infinite methods of adding details in terms of who should get alerts of what kind with what kind of escalation and how will resolution feedback be re-circulated (etc) 
 
Kaizen - The way in which Kaizens are identified, planned, executed and the way feedback and efficiency is tracked provides continuous improvement opportunities.
 
The Lean Champion needs to be on the lookout for methods and approaches that make Lean tools more effective and therein lies the opportunity of maximising the value generation for a Lean Enterprise.