A Julia module that implements the (normalized) iterative hard thresholding algorithm(IHT) of Blumensath and Davies. IHT performs feature selection akin to LASSO- or MCP-penalized regression using a greedy selection approach.
License: Other
IHT.jl implements iterative hard thresholding in a (decently) efficient manner. It provides crossvalidation routines to select the best sparsity level among a vector of possible levels. Finally, it interfaces with PLINK.jl to enable GWAS analysis.
Nonetheless, the facilities in IHT.jl are still somewhat limited and ripe for improvements and extensions. Among the proposed improvements are:
I have a dataset with 519 individuals and 10,000 SNPs.
I get a BoundsError when I try to run L0_log with k >= 3:
using DataFrames, IHT
dat = readtable("test.txt", header=true)
y = convert(Array{Float64,1},dat[:,1]) ##1st column: response
X = convert(Array{Float64,2},dat[:,2:end]) ##10,000 SNPs
output1 = @time L0_log(X, y, 1)
#0.004151 seconds (389 allocations: 1.165 MiB)
#IHT results:
#Compute time (sec): 0.0041
#Final loss: 0.6860874
#Iterations: 4
#IHT estimated 1 nonzero coefficients.
#1×2 DataFrames.DataFrame
#│ Row │ Predictor │ Estimated_β │
#├─────┼───────────┼─────────────┤
#│ 1 │ 8550 │ 0.331812 │
output2 = @time L0_log(X, y, 2)
#0.006286 seconds (445 allocations: 1.176 MiB)
#IHT results:
#Compute time (sec): 0.0062
#Final loss: 0.6840915
#Iterations: 4
#IHT estimated 2 nonzero coefficients.
#2×2 DataFrames.DataFrame
#│ Row │ Predictor │ Estimated_β │
#├─────┼───────────┼─────────────┤
#│ 1 │ 4307 │ 0.193256 │
#│ 2 │ 8550 │ 0.194642 │
output3 = @time L0_log(X, y, 3)
#ERROR: BoundsError: attempt to access 519×3 Array{Float64,2} at index [Base.OneTo(519), Base.OneTo(4)]
#Stacktrace:
# [1] throw_boundserror(::Array{Float64,2}, ::Tuple{Base.OneTo{Int64},Base.OneTo{Int64}}) at ./abstractarray.jl:433
# [2] checkbounds at ./abstractarray.jl:362 [inlined]
# [3] copy!(::Array{Float64,2}, ::SubArray{Float64,2,Array{Float64,2},Tuple{Base.Slice{Base.OneTo{Int64}},Array{Int64,1}},false}) at ./multidimensional.jl:803
# [4] #L0_log#90(::IHT.IHTLogVariables{Float64,Array{Float64,1}}, ::Float64, ::Float64, ::Float64, ::Float64, ::Float64, ::Int64, ::Int64, ::Bool, ::Bool, ::IHT.#L0_log, ::Array{Float64,2}, ::Array{Float64,1}, ::Int64) at /Users/Clauberry/.julia/v0.6/IHT/src/log.jl:343
# [5] L0_log(::Array{Float64,2}, ::Array{Float64,1}, ::Int64) at /Users/Clauberry/.julia/v0.6/IHT/src/log.jl:207I attach the test data file.
test.txt
The package cannot be installed with Pkg.clone("https://github.com/klkeys/IHT.jl.git") as it depends on two other packages that are not in METADATA. So, in order to install the package, one needs to do:
Pkg.clone("https://github.com/klkeys/PLINK.jl.git")
Pkg.clone("https://github.com/klkeys/RegressionTools.jl.git")
Pkg.clone("https://github.com/klkeys/IHT.jl.git")
in this order.
If you started with Pkg.clone("https://github.com/klkeys/IHT.jl.git"), you would need to remove the IHT folder in the .julia path, before trying again in the specific order above.
It would be nice to have this info in the documentation!
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
Django
The Web framework for perfectionists with deadlines.
Laravel
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft
Open source projects and samples from Microsoft.
Google
Google ❤️ Open Source for everyone.
Alibaba
Alibaba Open Source for everyone
D3
Data-Driven Documents codes.
Tencent
China tencent open source team.