Statistics is the field of data analysis and presentation. One essential phase in Statistics is data collection because it would reduce the complexity in analysis process and make the data easier to understand. Based on Applied Statistics and Probability for Engineers (Montgomery & Runger, 2003), three basic methods are considered: Retrospective study, observational study, and designed experiment.
Retrospective study, in general, collect historical data archived over given period of time. This method could result a significant amount of data for analysis; however, it is possible that only small portion of information appears from the data. A number of relationships between variables may be slightly invisible and some data may also be missing or improperly recorded which leads to difficulties in explaining the insights within the data.
Observational study obtains the data through inspection with little disruption possible. This method leaves the system in the current state as it is then record required data for analysis. This method requires relatively less time compared to other method since what the engineers would do is just observing and recording the present data.
Designed Experiment method is the opposite of observational study: Changes and adjustments are deliberately made as control variables then engineers collect the output data with respect to those changes. In other words, Engineers apply different treatments to the same sample of data, then collect the changes in output data. For example, an engineer observes how a difference in various machine settings affects the process time. From those data, engineers run statistical tests to determine which variable affect the changes. This method is impactful approach for complex system with multiple variables since it could show the changes in result from all possible variables combination.
Every method has its own strength and weaknesses depending on the situation. To decide which method to implement requires knowing which data to collect; hence, thorough assessment of the system is also crucial.
Comments