Choosing an Appropriate Distribution

We show how to create a Bernoulli that quantifies a binary discrete function.

We show how to create a Binomial that quantifies several binary discrete functions simultaneously.

We show how to create a Poisson to count discrete events with a single parameter.

We introduce the world of a much larger world of distribution functions: the continuous ones.

With availability of relevant data, we show how to use @RISK's tool to search for appropriate distributions.

With unavailable data we can use the opinion of an expert to ellicit an appropriate distribution.

We show how to create an Uniform distribution that usually typifies maximum uncertainty.

A common discussion among risk practitioners is explained here to understand both distribution differences and commonalities.

We show how to create an compare both a PERT and a Triangular distribution.

We show how to create an Triangular distribution, a very popular one in project management.

We show how to create Alternative distributions focused on their tails to create a better distribution.

Depending on the industry, many specialized distributions become popular ones based on their mathematical assumptions.

Whenever an expert can come up with the parameters of an industry specific distribution, its creation and use becomes straightforward and easy.

An example of the oil & gas industry which commonly uses lognormal distributions, in this case with percentile parameters.

We introduce general methodologies to combine several distributions to come up with creatively synthesized distributions.

We introduce the usage of kernel or synthesized and fitted distributions.

We combine PERT distributions and a Bernoulli to create a synthesized and fitted distribution that summarizes expert opinion.

We combine PERT distributions and a Discrete to create a synthesized and fitted distribution that summarizes expert opinion.

We show how to create a synthesized and fitted kernel distribution using 4 distributions together.

We introduce convolution, an important topic on risk quantification when considering frequency and severity parts of a risk model.

We show a first simplified approach on how to tackle down the problem of integrating frequency and severity components into a risk model.

We show a second manual method that allows for a better representation of the tails of a distribution when considering frequency severity models.

We finally introduce the COMPOUND function, a powerful @RISK function that adequately handles convolution.

When there is limited data, we present the functionality of this distribution that allows the interesting ellicitation of a continuous distribution.

With @RISK's Artist you can get really creative producing your own distributions.

A final note on the Normal and LogNormal distributions.