Yes, how do you feel the result?

On some areas it has cleaned up excessive aa and I enjoy the simplicity, overall it feels like trying to pixel with one hand tied behind my back. It certainly has a more 'pixelly' look, it's a bit low-fi for me though. It's satisfying to make clusters click, so it's probably a worthwhile exercise on less complicated images.

I will say that while it suffers a loss in fidelity, the actual pixel relationships are much more interesting.

Adjusted the palette (wider range of values with much darker darks and slightly darker midtones, also bumped up the saturation on the lighter colors)
Slimmed down the shoulders. Wide shoulders are a masculine trait.
Lengthened legs. Also adjusted leg anatomy. You're neglecting the calves and length of the tibia/fibula.
Brought the hat down onto her head. added drop shadow on face and hair from hat.
moved breast and defined the side of the torso.
defined the hands. made one ear floppy. gave the hat some character.

As jams0988 said, drawing the forms before adding clothing would be the best thing to do. Studying real human anatomy and proportions will also be of great benefit.

Also work on a neutral-colored background so you get a more non-biased idea of your color relationships.

Thanks Dennis. Your analysis is pretty accurate, it's an illustration of part of a poem I wrote:
Die and come back as a plant inside your house / I will live upon the air that you breathe out
Glad to see my intent was wasn't lost.

And might I say those are some mighty fine birds you've posted, all that practice is clearly paying off.

-->

used this as reference.
I think there are more interesting shapes in his face that you could highlight. What I would do differently is use less arbitrary restrictions like omitting AA and only using certain angles or palettes, and instead focus on filling the space with more cluster- and shape-interaction.

Not a very good cluster study imo, there really aren't that many cluster relationships to manage. Despite its size it's mostly giant clusters defining just the major features, anything that looks 'vectory' is probably not ideal for cluster focus. Think about more compact images that try to describe maximum detail in minimal space.

Example: (by thernz)