mschauer / CausalInference.jl

Causal inference, graphical models and structure learning in Julia

https://mschauer.github.io/CausalInference.jl/latest/

Other

189 stars 24 forks source link

CMI tests & tables #8

Closed schnirz closed 5 years ago

schnirz commented 5 years ago

implemented (conditional) independence tests based on nearest neighbor estimates of (conditional) mutual information
implemented interface for PC algorithm that accepts input data in table format and allows specification of independence test to use

All code for mutual information estimators is contained in a new file, klentropy.jl.

codecov-io commented 5 years ago

Codecov Report

Merging #8 into master will decrease coverage by 1.32%. The diff coverage is 65.34%.

@@            Coverage Diff             @@
##           master       #8      +/-   ##
==========================================
- Coverage   69.86%   68.54%   -1.33%     
==========================================
  Files           5        6       +1     
  Lines         292      391      +99     
==========================================
+ Hits          204      268      +64     
- Misses         88      123      +35

Impacted Files	Coverage Δ
src/pc.jl	`71.42% <46.66%> (-4.09%)`	:arrow_down:
src/klentropy.jl	`66.66% <66.66%> (ø)`
src/skeleton.jl	`68.42% <87.5%> (+2.42%)`	:arrow_up:

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update a31ef1c...b881246. Read the comment docs.

coveralls commented 5 years ago

Coverage decreased (-19.5%) to 50.386% when pulling 7df8a131e3381bd3632be63b74768289f4d89cf7 on schnirz:cmi_tests into a31ef1c07a78de1935f834ada04ef0ff9eae5254 on mschauer:master.

coveralls commented 5 years ago

Coverage decreased (-19.5%) to 50.386% when pulling 7df8a131e3381bd3632be63b74768289f4d89cf7 on schnirz:cmi_tests into a31ef1c07a78de1935f834ada04ef0ff9eae5254 on mschauer:master.

coveralls commented 5 years ago

Coverage decreased (-19.5%) to 50.386% when pulling 7df8a131e3381bd3632be63b74768289f4d89cf7 on schnirz:cmi_tests into a31ef1c07a78de1935f834ada04ef0ff9eae5254 on mschauer:master.

coveralls commented 5 years ago

Coverage decreased (-19.5%) to 50.386% when pulling 7df8a131e3381bd3632be63b74768289f4d89cf7 on schnirz:cmi_tests into a31ef1c07a78de1935f834ada04ef0ff9eae5254 on mschauer:master.

coveralls commented 5 years ago

Coverage decreased (-19.5%) to 50.386% when pulling 7df8a131e3381bd3632be63b74768289f4d89cf7 on schnirz:cmi_tests into a31ef1c07a78de1935f834ada04ef0ff9eae5254 on mschauer:master.

coveralls commented 5 years ago

Coverage decreased (-1.3%) to 68.542% when pulling b881246874256d3bdfced33d826cfd15a020656f on schnirz:cmi_tests into a31ef1c07a78de1935f834ada04ef0ff9eae5254 on mschauer:master.

schnirz commented 5 years ago

I suppose I should have seen this coming...I can start working on unit tests for the entropy estimators to get these numbers up again

mschauer commented 5 years ago

Great! Some remarks

yes, tests would be good
could you reduce the number of empty lines a bit?

use type assertions sparingly. Especially, do not make return type assertions, you write

function n_ball(n::Number)::Float64
 return π^(n/2.) / gamma(n/2. + 1.)
end

This makes

julia> ForwardDiff.derivative(n_ball, 10)
ERROR: MethodError: no method matching Float64(::ForwardDiff.Dual{ForwardDiff.Tag{typeof(n_ball),Int64},Float64,1})

fail for no reason. I'd rather have

julia> function n_ball(n::Number) return π^(n/2.) / gamma(n/2. + 1.) end n_ball (generic function with 1 method)

julia> ForwardDiff.derivative(n_ball, 10) -0.7158154677840614