Conway–Maxwell–Poisson distribution

Conway–Maxwell–Poisson (CMP) distribution is a generalized version of the Poisson distribution, which is widely used in statistics for modeling count data. The CMP distribution is particularly useful in situations where the data exhibit over-dispersion or under-dispersion relative to the Poisson distribution. This flexibility makes it applicable in various fields such as insurance mathematics, queueing theory, and biostatistics.

Definition

The probability mass function (PMF) of the Conway–Maxwell–Poisson distribution for a random variable X taking integer values is given by:

\[P(X = k) = \frac{\lambda^k}{(k!)^\nu Z(\lambda, \nu)},\]

where:

• \(\lambda > 0\) is the rate parameter,
• \(\nu \geq 0\) is the dispersion parameter,
• \(Z(\lambda, \nu) = \sum_{k=0}^{\infty} \frac{\lambda^k}{(k!)^\nu}\) is the normalization constant ensuring that the probabilities sum up to 1.

The parameter \(\nu\) controls the dispersion of the distribution. When \(\nu = 1\), the CMP distribution simplifies to the standard Poisson distribution. Values of \(\nu < 1\) indicate over-dispersion (variance greater than the mean), while \(\nu > 1\) indicate under-dispersion (variance less than the mean).

Properties

Mean and Variance

The mean and variance of the CMP distribution are not available in closed form and generally require numerical methods for their computation. However, these properties are crucial for understanding the behavior of the distribution and for fitting it to data.

Special Cases

• When \(\nu = 1\), as mentioned, the CMP distribution becomes the Poisson distribution.
• When \(\lambda = 1\) and \(\nu = 0\), it simplifies to the geometric distribution.

Applications

The Conway–Maxwell–Poisson distribution has been applied in various domains:

• In queueing theory, it is used to model the number of arrivals or services that occur in a given time period, especially when these events do not follow the assumptions of the Poisson distribution.
• In biostatistics, it can model count data such as the number of occurrences of a particular event within a fixed period or space, accommodating varying levels of dispersion.
• In insurance mathematics, the CMP distribution helps in modeling claim counts, providing a more flexible framework than the Poisson distribution for capturing the variability in claim frequencies.

Fitting the CMP Distribution

Fitting the CMP distribution to data involves estimating the parameters \(\lambda\) and \(\nu\). This can be achieved through maximum likelihood estimation (MLE), which requires numerical optimization techniques due to the lack of closed-form solutions for the parameters.

See Also

• Poisson distribution
• Dispersion (statistics)
• Geometric distribution
• Queueing theory
• Biostatistics