Add accessor functions

[ararslan]

Currently to get the t-statistic and degrees of freedom from a OneSampleTTest, you need to do OneSampleTTest(x).t and OneSampleTTest(x).df, respectively. It seems to me that it would be nice to provide a unified interface for extracting the test statistic and degrees of freedom from a HypothesisTest object. I was thinking of something along the lines of:

statistic(OneSampleTTest(x)) == OneSampleTTest(x).t
dof(OneSampleTTest(x)) == OneSampleTTest(x).df

where dof would be extended from StatsBase.

Thoughts?

[DominiqueMakowski]

Any advances for easing the access to test's parameters?

[nalimilan]

Probably not. Help welcome.

[pdeffebach]

It seems like every statistic reported in the table of output should have an accessor method. Would be nice to go through all the show_params methods for each test and add a function for each.

[StevenSiew]

Why not return a dictionary. Then the user can query the dictionary for the parameter they are interested in.

mydict = getparamdict(OneSampleTTest(x))
# show the output
println(  mydict["output"]  )
# show the value of t
println("t is ", mydict["t"]  )

[nalimilan]

Building a dictionary is quite inefficient when you just need one value.

[ararslan]

For the multivariate tests I ported over from MultivariateTests.jl, I actually did add the dof extension, so this feature request is now about 5% implemented.

[pdeffebach]

I wouldn't mind handling this. I can just go through and add all the things that are listed in the show_params function for each test.