0

I wonder if you can help me with this.

I've got homework where I need to determine the type of probability distribution that I should use to model the variable 'time'.

The variable however is measured as days in 0.5 day intervals. Would this be a discrete variable then even though time is usually a continuous variable?

If so, which distribution (Poisson, etc.) would be appropriate to model this?

Thank you

2 Answers2

0

If data are given so. You treat variables as dictreete. Unless you have to aproximate missing data in whole timespan

0

Yes, if all the possible values of your random variable are multiples of a given number (such as $0.5$), the random variable is discrete.

Without further information, it's impossible to say what the actual distribution would be. It might not be a "named" distribution. Note that the usual "named" discrete distributions take values that are integers, not half-integers. But it's possible that twice your random variable, which would take integer values, has a "named" distribution, perhaps geometric or negative binomial.

Robert Israel
  • 448,999