Most predictions are wildly optimistic.
5000 members after one year, 15000 members after two years. Even if you could get this many registered members, only a handful will be active.
The overwhelming majority of communities we consider successful have around 100 active members. That's all.
More accurate predictions are usually in the 50, 500, 1000 region. That's 50 after a month, 500 after a year, 1000 after two years.
There are better, more accurate, ways of making predictions. This all depends upon whether you have a mature community or an inception-stage community.
1) Current growth rate. All things considered, your growth rate today is likely to be the growth rate tomorrow. Next week is likely to be similar to last week. For mature communities, last year is likely to be similar to next year. Simply by knowing your current growth rate, you have a rough estimate for future growth rates.
2) Growth of similar communities. Speak to people who have founded similar communities. How many members did they have after 1 week/month/year? Be careful not to select solely the hyper-successful communities here. You need an average of several similar communities (the good and the bad).
3) Existing networks. If you're just starting, how many people do you personally know well enough to invite to join the community? How many people do they know? Test the basic reproduction number (R0) with a free, sharable, resource. This is a term to measure the likely spread of diseases. This will give you an idea of how many people will tell others about the community. Now you have a rough idea of the potential growth.
If you're really clever, you can put together a simple model which combines elements of the above to give you an idea of the future growth of the community.
We need to be better at making growth predictions. Bad predictions, even made in good faith, set high expectations that reflect badly on us in the long-run. Far better to take the time to make a realistic prediction.