Fitting Custom Distributions to Data Using JMP(R) (2019-US-45MP-272)
Aug 27, 2019 12:46 PM
| Last Modified: Nov 19, 2019 12:06 PM
Distribution talk Discovery 2019.zip
Matthew Flynn, Senior Data Scientist, GM Mary Loveless, JMP Systems Engineer, SAS
Not everything in the statistical world is normally distributed. For example, JMP includes the standard Poisson, Gamma Poisson (or Negative Binomial), Binomial and Bet Binomial distributions for fits for discrete data. However, the Nonlinear platform enables the fitting of your choice of distribution. The right “fit” is important for style, but it is also important in analytics to most accurately summarize one’s data as concisely and accurately as possible.
The JMP Distribution platform includes both commonly used discrete fit and continuous distributions. On occasion we have data with more unusual characteristics that require different distribution fits. How would you go about fitting and plotting new, custom distribution to this type data? In this presentation we will show how JMP makes it easy to work with alternative distributions, such as Simplex, L-Logistic and Kumaraswamy distributions for bounded responses; or alternative count data distributions, such as COM-Poisson, Consul, Double Poisson and a large number of discretized continuous distributions like the discrete Weibull or discrete Lindley distributions.