Anita Upadhyay

New generation tools and techniques allows us to measure practically everything and it is cheap to do so. This means that measure and analyze phase of DMAIC will no longer be impacted due to lack of sufficient data. Predictive modelling can be used in analyze and improve phase to make it more effective since the data availability is no longer a constraint. It will now become increasingly important to select relevant data from mountains of information available. It is not Lean Six Sigma vs BigData but BigData is an answer to a critical problem that plagued Six Sigma since its advent about data availability and analytics.
Descriptive Analytics – Insight into Past
Descriptive analytics is a preliminary stage of data processing that creates summary of historical data to yield useful information and possibly prepare for further analysis. It uses data aggregation and data mining to provide insights on what has happened.
Descriptive Statistics is a method of organizing, summarizing, and presenting data in a convenient and informative way with aim of understanding what has happened or current situation and aids in descriptive analytics. The actual method used depends on what information we would like to extract. The tools and techniques covered in Six Sigma that are applied in Descriptive analytics are measures of central tendency like mean, median, mode and quartiles and measures of dispersion / variation like standard deviation, variance and range. This is reasonably well captured in Lean Six Sigma.
Predictive Analytics – Understanding the future
Predictive analytics uses statistical models and forecast techniques to understand the future and answer what could happen. It helps predict what could happen based on data and these predictions are not 100% certain and is this uncertainty is denoted in form of probability. It uses historical data available within organizations to identify patterns and apply statistical models to forecast customer behavior, purchase patterns, inventory and sales. Another common application is to compute credit score. It helps fill in information that is not available based on information available. The tools and techniques covered in Six Sigma that can be applied in Predictive Analytics are Hypothesis testing, correlation and regression.  This is also reasonably well captured in Lean Six Sigma.
Prescriptive Analytics – Advise on possible outcomes and how to influence it
Prescriptive Analytics is a relatively new field that allows users to “prescribe” a no. of possible actions and guide them towards a solution. It helps to quantify effect of future decisions in order to advice on possible outcomes before actual decision is made. So it provides not just insight on what will happen but why it will happen in terms of actions that are required to ensure prediction is realized and how best to maximize benefits. It uses a combination of techniques and tools like business rules, algorithm, machine learning and computational modelling procedures which are applied against input from various sources like transactional, historical data, real time data feeds and big data. These are relatively complex and most companies aren’t applying it yet. The tools and techniques covered in Six Sigma that can be applied in Prescriptive Analytics are Design of Experiments and Simulation. This is captured in Lean Six Sigma however can be handled better with Prescriptive Analytics using BigData.

Cause is a variable which affects the outcome by increasing its chances or making it happen.
Problem is the outcome of the cause.
Scenario
Understanding  
Conclusion
Action
A cause (X) may be necessary but not sufficient for a problem (Y) to occur.
Problem (Y) occurs ONLY when cause (X) is TRUE
Problem (Y) may not occur even when the cause (X) is TRUE
Causes other than X (A, B, …) must occur together with X for problem Y to occur
Fix the cause (X) to remove one of the factors causing the problem (Y) however this will not fix it so look for other causes (A, B, ..) that result in this problem (Y) and address them too to fix the problem (Y).
A cause (X) may be sufficient but not necessary for a problem (Y) to happen.
Problem (Y) occurs whenever cause (X) is TRUE
Problem (Y) can occur even when cause (X) is not TRUE
Cause X is a definite cause for the problem (Y), however other causes (A, B, … ) can also result in this problem (Y)
Fix the cause (X) to address some scenarios in which the problem (Y) occurs.
However look for other causes (A, B, ..) which may also result in this problem to fix it for all scenarios.
A cause (X) may be neither sufficient nor necessary for a problem (Y) to occur.
Problem (Y) may not occur even when cause (X) is TRUE
Problem (Y) can occur even when cause (X) is not TRUE
Cause X is not established as a reason for the problem (Y)
Fixing cause (X) may not fix the problem (Y). Look for other causes (A, B, ..) which are likely to be more relevant cause for this problem (Y)
A cause may be both sufficient and necessary for a problem to occur.
Problem (Y) occurs ONLY when cause (X) is TRUE
Problem (Y) occurs whenever cause (X) is TRUE
Cause X is a definite cause for problem Y
Fixing cause (X) may fix Y completely.