Use of Six sigma tools on real time data
Mohit Sharma
Country Quality Head at Keysight Technologies |Certified Master Black Belt & Expert in RPA implementation| Certified Lead Auditor ISO 9001: 2015 | Scrum Coach
Six sigma tools like Pareto, Box plot, control chart and some of the basic statistics could be used to analyze real time data and situations.
Let us take some tools and scenarios to understand their usage
Pareto Chart – Pareto chart is created when data is discrete and if top 20% contributors of 80% of the problems are to be identified (it is also called 80/20 rule) for example
Scenario 1: There are errors happening in the process and they are categorized in 5 different categories. Training department is trying to create a training plan to impart knowledge. If data stratification is not done then they will train participants on all categories.This could be lengthy training plan. However if a Pareto is created and 80% contributing error categories are identified, then a small training on identified category can yield big result. see the table below.
Now look at the Pareto
78.6% errors are contributed by Spelling mistakes, Punctuation errors and Salutation errors. So the training on these 3 topics will cater for 78.6% of the problem. A shorter training in less time can tackle 78.6% errors.
Basic Statistics:
Scenario 2: In a newly transitioned transaction based process, number of Full time employees are to be calculated. Total no of transactions and time taken to process these transactions are given. Look at the table below to understand calculations
Basic statics can be used in the example above. Use of statistically validated sample size to calculate 'time taken' (minutes) per transaction. Sample calculator can be used or a thumb rule from central limit theorem can be refereed to where a minimum sample of 30 samples per subgroup are advised to be considered.
While calculating the 'time taken', check the normality of the time data collected (as per the above sample) and if the data is normally distributed then use mean and if it is non normally distributed then use median to represent central tendency.
If the above recommended method would be used then the accuracy of the FTE calculation would be close to real.
Hence in this way six sigma tools can be used on real time data.
Project Manager at Praxma Consulting
8 年Thanks Mohit, sometimes is better go back to basics !