Tomi Ahonen wrote a blog post about the migration from feature (or dumb) phones to smartphones:
The typical diffusion curve is not in the shape of a straight line, it is rather an S-shaped curve, and the final migration rates should slow down, meaning we would push the 99% level into year 2021 or even 2022.
–Tomi Ahonen, October 19, 2017 
I noted that there started to be disagreeing comments that highlighted Tomi has previously given a slightly different forecast that gives 100% migration rate a lot earlier:
And the migration from dumbphones to all new phones sold being smartphones I have happening around early 2019.
–Tomi Ahonen, July 23, 2014 
the migration rate from dumbphones to smartphones (we should be 100% migration completed by year 2019 by current projections).
–Tomi Ahonen, February 24, 2015 
That market will evolve to become 100% smartphones just like how black-and-white TV sets evolved to color TV sets, and CRT-based ‘boxy’ style old TVs evolved to flat panel TV sets. It is an unstoppable train, the dumbphone market will end, it is expected to end by 2020 (I have been saying end of 2019).
–Tomi Ahonen, April 29, 2016 
The iPhone will continue to have great sales success and a MODEST market share which will be near 10% when all new phones sold, will be smartphones around the end of year 2019.
–Tomi Ahonen, May 04, 2016 
Tomi will (when he no longer cannot just delete the comments) blame Stephen Elop for his failed forecast.
Yes, Tomi will say that Elop blew Nokia migration to smartphones, which turned smartphone users back to feature phone users, thus reversing the trend and causing Tomi’s forecast to fail.
(Please ignore the fact Elop “did that” in 2012, Tomi made his forecast somewhere in late 2013 or early 2014 and was still using 2019 deadline in May 2016.)
It’s 20:29 GMT. If Tomi makes his stand even a minute later, you heard it here first.