Proper data collection is very important to determine where we are and where we want to go in our Six Sigma project. The precision, consistency, and strategies employed during the collection of data has a large bearing on our ability to accurately analyze, improve, and control.....After all, garbage in -- garbage out.
The majority of data collected is never looked at. And because collecting data adds no real value to the product or service delivered, the gathering of this data is a Non-value added step. We should only collect the data that we intend to use, and not waste time with not-useful data gathering.
There are not too many concepts that touch every phase of a Six Sigma project. One that does is collecting data...
In the Define phase, data is collected to identify and define projects. It’s also required to define the projects primary and consequential metrics. These metrics track the success of the project as it move through DMAIC process.
In the Measure phase, data collection is required to help assess the measurement system of the project metrics that we’re tracking. We also use this data to understand the capability of the processes through capability analysis.
In the Analyze phase, we use either historical data or collect data while the process is runs in its standard mode to analyze this information in a passive manner. Analyzing passive data, helps us identify sources of variation that can contribute to the problems we’re seeing in the process.
In the Improve phase instead of passive data collection, a data plan is needed to manage the experimental changes that are made in the process seeking improvement. Data will need to be collected it while we’re actually changing the process in a very deliberate way.
In the Control phase, we use the data collected and analysed from the Define, Measure, Analyze and Improve phases to create structured control plans. Those control plans will define the data collection on the X’s or Y’s that are meaningful to managing the process long term.
There are two main groups of data, Continuous Data and Discrete Data.
Continuous data is data from a measurement scale that can be divided into finer and finer increments. (It’s also called Variable Data.) For example: measurements of weight, speed, temperature, length are all examples of this type of data.
This type of data is very information-rich, meaning small numbers of samples can provide large amounts of information. It is the data of choice, always use variable data when you can.
Lastly there are many, many more statistical tools available for extracting this type of data.
The other type of data is discrete data. It is also called attribute data. Discrete data is count data associated with some group or category. For example: pass or fail data, yes or no data, present or not present data, scale of 1 to 10 data. All of these are discrete data.
Discrete data is information-poor. Many samples are required to see differences. There are fewer statistical tools available for extracting information from this type of data. Whenever possible we should use continuous data instead of discrete data.
Learn more about types of data here.
One primary principle in data collection is obtaining representative samples. When collecting data, we must always ask ourselves if our example is representative. The idea is to have as few samples as possible while still providing accurate depictions of the entire population.
Different sampling strategies exist to help ensure the collection of representative samples. The most common of these sampling methods is called simple random sampling.
Samples need to be selected at random to increase the likelihood that the measurements of the samples are representative of the process population. It improves the validity of population estimates and statistical tests such as capability analysis or the calculation of the probability of defect. Sampling only the first few parts does not show how the process performs over time.
Manual methods are still a quick way of gathering data. Individuals working within the process implement the data collection plan at the appropriate place in the process. Typical approaches use a standard template with row and column labels. Another effective method is the check sheet.
The preferred manner of data gathering is obviously through automated means. In the ideal case, a computer will listen to process characteristics and automatically gather the information for you. This information will be timely and accurate. Data summaries are faster to obtain and could take seconds rather than weeks.
May 10, 16 09:24 PM
A Quality Control Plan is a documented description of the activities needed to control a process or product. The objective of a QCP is to minimize variation.
May 10, 16 08:49 PM
The Largest Collection of Free Six Sigma Tools and Training on the Web!
May 10, 16 07:28 PM
The Weibull distribution is applicable to make population predictions around a wide variety of patterns of variation.