Understanding the different types of data and their respective uses is critical for product development, testing, and analysis. Each type of data plays a role in ensuring that products meet quality standards and fulfill user needs. As a mechanical engineer with a focus on R&D testing and data analysis, you would likely encounter and utilize these various data types throughout the product development and validation process.
Fitting data to a distribution is a critical step in statistical analysis because it allows for the development of models that accurately represent the underlying random processes of the data. When data is properly fitted to a distribution, it enables precise calculations and predictions, which are essential for making informed business decisions. Using an incorrect distribution can lead to erroneous conclusions, potentially resulting in significant time and financial losses, or even damage to equipment in engineering applications.
Data Collection and Preparation:
Gather the data from field tests or experiments.
- Record the exact time each subcomponent fails, along with any relevant circumstances or symptoms observed at the time of failure.
- Create histograms:In next graphs presented 2 sets of data from the time to failure of 2 different subcomponents.
Fit probability plots to visually assess the distribution of the data:
Use statistical methods to estimate the parameters of the candidate distributions.
Perform goodness-of-fit tests, such as the Kolmogorov-Smirnov test, Anderson-Darling test, or chi-square test, to determine how well the distribution fits the data. I chose relay on AD (Anderson-Darling) fit factor:
The Anderson-Darling method is a statistical test used to determine if a sample of data comes from a population with a specific distribution. It is particularly sensitive to deviations in the tails of the distribution, making it more effective for identifying departures from normality.
A lower Anderson-Darling (AD) statistic indicates a better fit between the sample data and the specified distribution. The AD test is a measure of the distance between the empirical distribution function of the sample data and the cumulative distribution function (CDF) of the specified theoretical distribution. A smaller value of the AD statistic means that the sample data more closely follows the theoretical distribution, suggesting a good fit.
Distribution fit models are essential tools in predictive analytics, as they allow analysts to understand the underlying patterns and trends within historical data. By fitting a statistical distribution to data points, analysts can make informed predictions about future events or behaviors. This process involves selecting the most appropriate distribution model that best represents the data’s characteristics, such as normal, exponential, or log-normal distributions, among others.
One of the use is to define time when the 10% of population will be in danger of failure:
Have more questions? Please lets discuss in comments section!
Leave a Reply