Variation is a lack of consistency in the output of our process. So, for example, we say that we made something in 30 days and sometimes we need 25, sometimes 30, sometimes 35.


A slight variation can be caused by common causes already present in the process. On the other hand, special causes of variation can be caused by a problem that we need to investigate and fix. 


We need to fix the special causes of variation because they cause:

  • Loss of money, when something doesn’t meet the customer specification;
  • Reliability, because the customer aspect from us some specification level or service level.


In every process, we have this 6M that are different element that can generate variation:

  • Method
  • Mother Nature, or Environmental 
  • Man 
  • Measurement 
  • Machine 
  • Materials

Variation can occur in each one of these. 

In addition we can have difference source of variation:

  • Within piece: variations that occur in a single piece;
  • Piece to piece: variations that occur between different piece;
  • Time to time: variations that occur at different times;

Variation is about the quality of our process, and it can be measured by:


But what’s the multi-vari analysis?

It’s about breaking down the variation to look at the different sources of it. In matemathics notation is when you have more dependant variable Y1,..,Yn and you want to look at multiple indipendant variable X1,..,Xn that affect they.


You can look an example of multi-vari analysis in the chapter 4.2.2 Multiple Linear Regression.


Another example of multi-vari analysis is the multi-vari chart, which visually presents variability through a series of charts. In image1 we have an example of one of the possible form of the multi-vari chart.

Image1 - Muti-vari chart example
Image1 – Muti-vari chart example

In image1 we have:

  • The dependent variable Y that is the temperature;
  • The independent variables x1 that are the operator (with value 1,2 and 3) and x2 that is the measure (with values 1,2 and 3 that is a circle, triangle, and rhombus);
  • The mean of each operator, that is represented with the square.


Example: in image1, we have three different operators who read the measure of temperature from 3 separate sensors (measure1, measure2, and measure3). We see that sensor2 (the triangle) always has a lower value for each operator with the multi-vari chart. Plus, you can easily see that operator 3 has a higher mean value than operator1 and operator2. So this chart suggests making some inspections on sensor2 and operator3.

References:

Share on: