normal distribution vs. poisson

Hello. I'm reading a book on creating random numbers. If I'm trying to come up with units per week with an average of 100, why would I use the poisson distribution vs. the normal distribution?


Item sales are in unit counts, so we use the Poisson distribution to generate countdata: rpois(n, lambda), where n is the number of draws and lambda is themean value of units per week. We draw a random Poisson count for each row(nrow(store.df), and set the mean sales (lambda) of Product 1 to be higherthan that of Product 2:

amartino44 said:

Hello. I'm reading a book on creating random numbers. If I'm trying to come up with units per week with an average of 100, why would I use the poisson distribution vs. the normal distribution?


Item sales are in unit counts, so we use the Poisson distribution to generate countdata: rpois(n, lambda), where n is the number of draws and lambda is themean value of units per week. We draw a random Poisson count for each row(nrow(store.df), and set the mean sales (lambda) of Product 1 to be higherthan that of Product 2:

Click to expand...

The Normal and Poisson distributions are appropriate under different circumstances. The Poisson distribution is generally more appropriate for a 'how many discrete (integer) things will occur during a fixed time interval' than the Normal distribution. One reason is that the Normal distribution would allow for -2.5 things to happen and the Poisson distribution only allows for non-negative integer things to happen.

thank ye

Got it. Thanks!!