GitHub

Robust regression

This example illustrates robust polynomial fitting with ℓₚ norm cost functions using the Julia language.

This page comes from a single Julia file: robust-regress.jl.

You can access the source code for such Julia documentation using the 'Edit on GitHub' link in the top right. You can view the corresponding notebook in nbviewer here: robust-regress.ipynb, or open it in binder here: robust-regress.ipynb.

Setup

Add the Julia packages used in this demo. Change false to true in the following code block if you are using any of the following packages for the first time.

if false
    import Pkg
    Pkg.add([
        "InteractiveUtils"
        "LaTeXStrings"
        "LinearAlgebra"
        "MIRTjim"
        "Optim"
        "Plots"
        "Random"
    ])
end

Tell Julia to use the following packages. Run Pkg.add() in the preceding code block first, if needed.

using InteractiveUtils: versioninfo
using LaTeXStrings
using LinearAlgebra: norm
using MIRTjim: prompt
using Optim: optimize
using Plots: default, plot, plot!, scatter, scatter!, savefig
using Random: seed!
default(); default(label="", markerstrokecolor=:auto, widen=true, linewidth=2,
    markersize = 6, tickfontsize=12, labelfontsize = 16, legendfontsize=14)

The following line is helpful when running this jl-file as a script; this way it will prompt user to hit a key after each image is displayed.

isinteractive() && prompt(:prompt);

Simulated data from latent nonlinear function

s = (t) -> atan(4*(t-0.5)) # nonlinear function

seed!(0) # seed rng
M = 12 # how many data points
tm = sort(rand(M)) # M random sample locations
y = s.(tm) + 0.1 * randn(M) # noisy samples
y[2] = 0.3 # simulate an outlier
y[M-2] = -0.3 # another outlier

t0 = range(0, 1, 101) # fine sampling for showing curve
xaxis = (L"t", (0,1), 0:0.5:1)
yaxis = (L"y", (-1.2, 1.7), -1:1)
p0 = scatter(tm, y; color=:black, label="y (data with outliers)",
 xaxis, yaxis)
plot!(t0, s.(t0), color=:blue, label="s(t) : latent signal", legend=:topleft)
Example block output
prompt()

Polynomial model

deg = 3 # polynomial degree
Afun = (tt) -> [t.^i for t in tt, i in 0:deg] # matrix of monomials
A = Afun(tm) # M × 4 matrix
p1 = plot(title="Columns of matrix A", xlabel=L"t", legend=:left)
for i in 0:deg
    plot!(p1, tm, A[:,i+1], marker=:circle, label = "A[:,$(i+1)]")
end
p1
Example block output
prompt()

LS estimation

This is not robust to the outliers.

xls = A \ y # backslash for LS solution using all M samples

p2 = deepcopy(p0)
plot!(p2, t0, Afun(t0)*xls, color=:magenta, label="LS fit")
Example block output
prompt()

Robust regression

Using (differentiable) p-norm with $1 < p ≪ 2$ avoids over-fitting the outlier data points.

p = 1.1 # close to ℓ₁
cost = x -> norm(A * x - y, p)
x0 = xls # initial guess
outp = optimize(cost, x0)
xlp = outp.minimizer

plot!(p2, t0, Afun(t0)*xlp, color=:green, line=:dash,
 label="Robust fit p=$p")
Example block output

Using 1-norm produces nearly the same results as using the p=1.1 norm.

cost1 = x -> norm(A * x - y, 1) # ℓ₁
out1 = optimize(cost1, x0)
xl1 = out1.minimizer

plot!(p2, t0, Afun(t0)*xl1, color=:orange, line=:dashdot,
 label="Robust fit p=1")

# savefig(p2, "robust-regress.pdf")
Example block output

Reproducibility

This page was generated with the following version of Julia:

using InteractiveUtils: versioninfo
io = IOBuffer(); versioninfo(io); split(String(take!(io)), '\n')
12-element Vector{SubString{String}}:
 "Julia Version 1.10.1"
 "Commit 7790d6f0641 (2024-02-13 20:41 UTC)"
 "Build Info:"
 "  Official https://julialang.org/ release"
 "Platform Info:"
 "  OS: Linux (x86_64-linux-gnu)"
 "  CPU: 4 × AMD EPYC 7763 64-Core Processor"
 "  WORD_SIZE: 64"
 "  LIBM: libopenlibm"
 "  LLVM: libLLVM-15.0.7 (ORCJIT, znver3)"
 "Threads: 1 default, 0 interactive, 1 GC (on 4 virtual cores)"
 ""

And with the following package versions

import Pkg; Pkg.status()
Status `~/work/book-la-demo/book-la-demo/docs/Project.toml`
  [6e4b80f9] BenchmarkTools v1.5.0
  [aaaa29a8] Clustering v0.15.7
  [35d6a980] ColorSchemes v3.24.0
  [3da002f7] ColorTypes v0.11.4
⌅ [c3611d14] ColorVectorSpace v0.9.10
  [717857b8] DSP v0.7.9
  [72c85766] Demos v0.1.0 `~/work/book-la-demo/book-la-demo`
  [e30172f5] Documenter v1.2.1
  [4f61f5a4] FFTViews v0.3.2
  [7a1cc6ca] FFTW v1.8.0
  [587475ba] Flux v0.14.12
⌅ [a09fc81d] ImageCore v0.9.4
  [71a99df6] ImagePhantoms v0.7.2
  [b964fa9f] LaTeXStrings v1.3.1
  [7031d0ef] LazyGrids v0.5.0
  [599c1a8e] LinearMapsAA v0.11.0
  [98b081ad] Literate v2.16.1
  [7035ae7a] MIRT v0.17.0
  [170b2178] MIRTjim v0.23.0
  [eb30cadb] MLDatasets v0.7.14
  [efe261a4] NFFT v0.13.3
  [6ef6ca0d] NMF v1.0.2
  [15e1cf62] NPZ v0.4.3
  [0b1bfda6] OneHotArrays v0.2.5
  [429524aa] Optim v1.9.2
  [91a5bcdd] Plots v1.40.1
  [f27b6e38] Polynomials v4.0.6
  [2913bbd2] StatsBase v0.34.2
  [d6d074c3] VideoIO v1.0.9
  [b77e0a4c] InteractiveUtils
  [37e2e46d] LinearAlgebra
  [44cfe95a] Pkg v1.10.0
  [9a3f8284] Random
Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated`

This page was generated using Literate.jl.