# Data considerations for Parametric Growth Curve

To ensure that your results are valid, consider the following guidelines when you collect data, perform the analysis, and interpret your results.

Collect failure times for a repairable system
System repair data usually consist of successive failure (or repair) times. For example, an automobile breaks down, is repaired, and returned to service, then breaks down again, and so on. The data values represent the time of each failure without considering the repair time.
Exact data can be failure truncated or time truncated
A failure-truncated system is retired after a certain number of failures occur. In a failure-truncated system, the system is retired immediately upon the final failure. A time-truncated system is retired after a specified period of time. In a time-truncated system, the largest time is not a failure time. If you have both failure-truncated data and time-truncated data, you must use a retirement column to indicate whether each time is a failure time or a retirement time. For more information, see "Specify retirement information".
The data can be exact failure times or failures within time intervals
If you have exact data, you know exactly when each failure occurred. For example, an engine failed at exactly 490 days, was repaired, then failed again at 822 days. If you have interval data, you know only that each failure occurred between two specific times. For example, an engine failed sometime between 475 and 500 days, was repaired, and then failed again sometime between 800 and 825 days.
Identify data from multiple systems
To evaluate multiple systems in a repairable system analysis, you must have a column that contains failure times and a corresponding column that identifies which system the failure is from. Minitab assumes that all the systems within a column are from identical processes and provides a pooled growth curve estimate. However, Minitab also tests for equal shapes or scales across the systems. If the test results indicate that the shape or scale for different systems are not equal, you should analyze the data from each system separately.
The model that you use must adequately fit the data
You can use a power-law process or a Poisson model for your data. Use a power-law process to model failure/repair times that occur at a rate that is increasing, decreasing, or constant. Use a Poisson model to model failure/repair times that occur at a rate that remains stable over time. To determine whether the model that you select adequately fits the data, use the plots and the trend tests. If the model that you select does not adequately fit your data, the results of the analysis may not be accurate. In that case, consider using Nonparametric Growth Curve.
By using this site you agree to the use of cookies for analytics and personalized content.  Read our policy