Hi everyone, I'm not sure what kinds of analysis models should be used for discrete percentages, here's an example:
the data were the number of counts, and also the percentages of the counts.
There were 10,000 baskets contained fruits (e.g. orange, apple, banana), and some of the baskets contained apples.
the number of apples (count), was the nominator
the number of baskets (count), was the denominator
so, the percentage is "number of apples / number of baskets".
is this kind of percentage referred to "discrete percentage"?
And the percentage can be very small (e.g. 0.01% of baskets contained apples).
I've tried using the GLM with linear or negative binomial distribution, although it returned significant result, the estimated parameters (e.g. Beta) were too close to 1 (e.g. 0.9998, or 1.0001).
I've also think about using time series model (e.g. ARIMA), as my data contained seasonality.
But any other analysis models that are suitable for the "discrete percentage"?
Thanks!