If you provide more information regarding the materials of construction (presumably some sort of hardened tool steel for the shaft, not sure about your collar) and actual dimensions (actual meaning you measured the parts with a micrometer, or other suitable measuring device, and not the nominal dimensions provided by the manufacturer or parts description) then you could use an online fit calculator such as this one provided by Engineers Edge to get an idea of the proper sizeinterference needed - maybe look to the FN5 class for example.
I'm suspecting you're going to need a significant interference to withstand the loads expected, so you'll probably need a hydraulic press or a good source of heat, or maybe a combination of the too.
Again, if you know the material of construction, you can use the coefficient of thermal expansion to determine the temperature change necessary to cause the female member to expand large enough to slip the male member in, and then when it cools and contracts it will create the interference necessary to hold the parts in place.
Another option you might consider, is that you can press fit the pieces together (I recommend the shrink fit method above to maintain concentricity) and since this is a splined shaft fitting into a round hole, you may be able to braze the shaft to the coupling (filling in the gaps in the teeth with brazing material) which would be extremely strong. Again without knowing more about the base materials I don't know if this is a suitable suggestion, but it's worth considering. Many cutting tools used for milling operations which have carbide inserts are brazed, and it's proven to be a reliable and strong connection for the job.
Standard disclosure: you should consult an engineer experienced with this type of application (someone specializing in machine design, or machine maintenance might be a good place to start) before just taking numbers off some website or the words of some guy who doesn't have all the information about the application. For one - the numbers suggested on the site I linked to are based off an ANSI standard that provides reccomendationsrecommendations for common applications of holding a round pin in a round hole - none of which describes your application. And for two - you're playing with some pretty large forces and heavy loads in a milling operation, and a catastrophic failure could send an extremely sharp large chunk of carbide, cobalt, or high-speed-steel straight at your face, eye, throat, or child-rearing organs which would not be very fun at all.