v0.9.0

Author: jbytecode

CHANGELOG.md

@@ -1,5 +1,10 @@
-# Upcoming Release 
+# Upcoming Release 
+
+
+# 0.9.0
 - Add exact argument for LAD. If exact is true then the linear programming based exact solution is found. Otherwise, a GA based search is performed to yield approximate solutions. 
+- Remove dependency of Plots.jl. If Plots.jl is installed and loaded manually, the functionality that uses Plot is autmatically loaded by Requires.jl. Affected functions are `dataimage`, `mveltsplot`, and `bchplot`.
+
 
 # v0.8.19
 - Update Satman(2013) algorithm

Project.toml

@@ -1,7 +1,7 @@
 name = "LinRegOutliers"
 uuid = "6d4de0fb-32d9-4c65-aac1-cc9ed8b94b1a"
 authors = ["Mehmet Hakan Satman <mhsatman@gmail.com>", "Shreesh Adiga <16567adigashreesh@gmail.com>", "Guillermo Angeris <angeris@stanford.edu>", "Emre Akadal <emre.akadal@istanbul.edu.tr>"]
-version = "0.8.19"
+version = "0.9.0"
 
 [deps]
 Clustering = "aaaa29a8-35af-508c-8bc3-b662a17a0fe5"
@@ -11,7 +11,7 @@ Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"
 GLPK = "60bf3e95-4087-53dc-ae20-288a0d20c6a6"
 JuMP = "4076af6c-e467-56ae-b986-b466b2749572"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
-Plots = "91a5bcdd-55d7-5caf-9e0b-520d859cae80"
+Requires = "ae029012-a4dd-5104-9daa-d747884805df"
 StatsModels = "3eaba693-59b7-5ba5-a881-562e759f1c8d"
 
 [compat]
@@ -21,6 +21,6 @@ DataFrames = "0.22, 1"
 Distributions = "0.17, 0.18, 0.19.1, 0.20, 0.21, 0.22, 0.23, 0.24, 0.25"
 GLPK = "0.10, 0.11, 0.12, 0.13, 0.14, 0.15, 1"
 JuMP = "0.19, 0.20, 0.21, 0.22, 0.23, 1"
-Plots = "0.22.2, 0.23, 0.24, 0.25, 0.26, 0.27, 0.28, 0.29, 1"
+Requires = "1"
 StatsModels = "0.4, 0.5, 0.6"
 julia = "1.4"

src/LinRegOutliers.jl

@@ -1,5 +1,26 @@
 module LinRegOutliers
 
+using Requires
+
+# After the module is loaded, we check if Plots is installed and loaded.
+# If Plots is installed and loaded, we load the corresponding modules.
+function __init__()
+    @require Plots = "91a5bcdd-55d7-5caf-9e0b-520d859cae80" begin
+  
+        import .Plots: RGB
+
+        include("mveltsplot.jl")
+        include("dataimage.jl")
+        include("bchplot.jl")
+
+        import .MVELTSPlot: mveltsplot
+        import .DataImage: dataimage
+        import .BCHPlot: bchplot
+        
+        export mveltsplot, dataimage, bchplot, RGB
+
+    end
+end
 
 # Basis
 include("basis.jl")
@@ -91,13 +112,14 @@ import .ASM2000: asm2000
 include("mve.jl")
 import .MVE: mve, mcd
 
+# Moved into grahhics.jl
 # MVE - LTS plot for visual detection of regression outliers
-include("mveltsplot.jl")
-import .MVELTSPlot: mveltsplot
+#include("mveltsplot.jl")
+#import .MVELTSPlot: mveltsplot
 
 # Billor & Chatterjee & Hadi Algorithm for detecting outliers
 include("bch.jl")
-import .BCH: bch, bchplot
+import .BCH: bch
 
 # Pena & Yohai (1995) algorithm
 include("py95.jl")
@@ -138,9 +160,10 @@ include("hadi1994.jl")
 import .Hadi94: hadi1994
 
 
+# Moved into graphics.jl
 # Gray-scale images of distance matrices
-include("dataimage.jl")
-import .DataImage: dataimage
+#include("dataimage.jl")
+#import .DataImage: dataimage
 
 
 # Modified and original Satman (2012) algorithms
@@ -218,8 +241,7 @@ export asm2000
 export lms
 export lts
 export mve, mcd
-export mveltsplot
-export bch, bchplot
+export bch
 export py95, py95SuspectedObservations
 export satman2013
 export satman2015, dominates
@@ -228,7 +250,6 @@ export quantileregression
 export lta
 export hadi1992
 export hadi1994
-export dataimage
 export gwcga, galts, ga, cga, RealChromosome
 export detectOutliers
 export ransac

src/bch.jl

@@ -1,7 +1,7 @@
 module BCH
 
 export bch
-export bchplot
+
 
 import ..Basis:
     RegressionSetting, @extractRegressionSetting, designMatrix, responseVector, applyColumns
@@ -13,8 +13,6 @@ import Distributions: Chisq
 import LinearAlgebra: diag
 import DataFrames: DataFrame
 
-using Plots
-
 
 
 """
@@ -207,59 +205,5 @@ end
 
 
 
-"""
-
-    bchplot(setting::RegressionSetting; alpha=0.05, maxiter=1000, epsilon=0.00001)
-
-Perform the Billor & Chatterjee & Hadi (2006) algorithm and generates outlier plot 
-for the given regression setting.
-
-# Arguments
-- `setting::RegressionSetting`: RegressionSetting object with a formula and dataset.
-- `alpha::Float64`: Optional argument of the probability of rejecting the null hypothesis.
-- `maxiter::Int`: Maximum number of iterations for calculating iterative weighted least squares estimates.
-- `epsilon::Float64`: Accuracy for determining convergency.
-
-
-# References
-Billor, Nedret, Samprit Chatterjee, and Ali S. Hadi. "A re-weighted least squares method 
-for robust regression estimation." American journal of mathematical and management sciences 26.3-4 (2006): 229-252.
-"""
-function bchplot(
-    setting::RegressionSetting;
-    alpha = 0.05,
-    maxiter = 1000,
-    epsilon = 0.00001,
-)
-    X = designMatrix(setting)
-    y = responseVector(setting)
-    return bchplot(X, y, alpha = alpha, maxiter = maxiter, epsilon = epsilon)
-end
-
-function bchplot(
-    Xdesign::Array{Float64,2},
-    y::Array{Float64,1};
-    alpha = 0.05,
-    maxiter = 1000,
-    epsilon = 0.00001,
-)
-    result = bch(Xdesign, y, alpha = alpha, maxiter = maxiter, epsilon = epsilon)
-    squared_normalized_residuals = result["squared.normalized.residuals"]
-    squared_normalized_robust_distances = result["squared.normalized.robust.distances"]
-    n = length(squared_normalized_robust_distances)
-    scplot = scatter(
-        squared_normalized_robust_distances,
-        squared_normalized_residuals,
-        legend = false,
-        series_annotations = text.(1:n, :bottom),
-        tickfont = font(10),
-        guidefont = font(10),
-        labelfont = font(10),
-    )
-    title!("Billor & Chatterjee & Hadi Plot")
-    xlabel!("Squared Normalized Robust Distances")
-    ylabel!("Squared Normalized Residuals")
-end
-
 
 end # end of module BCH 

src/bchplot.jl

@@ -0,0 +1,69 @@
+module BCHPlot 
+
+
+export bchplot
+
+
+import ..BCH: bch
+import ..Basis: RegressionSetting
+
+
+"""
+
+    bchplot(setting::RegressionSetting; alpha=0.05, maxiter=1000, epsilon=0.00001)
+
+Perform the Billor & Chatterjee & Hadi (2006) algorithm and generates outlier plot 
+for the given regression setting.
+
+# Arguments
+- `setting::RegressionSetting`: RegressionSetting object with a formula and dataset.
+- `alpha::Float64`: Optional argument of the probability of rejecting the null hypothesis.
+- `maxiter::Int`: Maximum number of iterations for calculating iterative weighted least squares estimates.
+- `epsilon::Float64`: Accuracy for determining convergency.
+
+
+# References
+Billor, Nedret, Samprit Chatterjee, and Ali S. Hadi. "A re-weighted least squares method 
+for robust regression estimation." American journal of mathematical and management sciences 26.3-4 (2006): 229-252.
+
+!!! warning "Dependencies"
+    This method is enabled when the Plots package is installed and loaded.
+        
+"""
+function bchplot(
+    setting::RegressionSetting;
+    alpha = 0.05,
+    maxiter = 1000,
+    epsilon = 0.00001,
+)
+    X = designMatrix(setting)
+    y = responseVector(setting)
+    return bchplot(X, y, alpha = alpha, maxiter = maxiter, epsilon = epsilon)
+end
+
+function bchplot(
+    Xdesign::Array{Float64,2},
+    y::Array{Float64,1};
+    alpha = 0.05,
+    maxiter = 1000,
+    epsilon = 0.00001,
+)
+    result = bch(Xdesign, y, alpha = alpha, maxiter = maxiter, epsilon = epsilon)
+    squared_normalized_residuals = result["squared.normalized.residuals"]
+    squared_normalized_robust_distances = result["squared.normalized.robust.distances"]
+    n = length(squared_normalized_robust_distances)
+    scplot = scatter(
+        squared_normalized_robust_distances,
+        squared_normalized_residuals,
+        legend = false,
+        series_annotations = text.(1:n, :bottom),
+        tickfont = font(10),
+        guidefont = font(10),
+        labelfont = font(10),
+    )
+    title!("Billor & Chatterjee & Hadi Plot")
+    xlabel!("Squared Normalized Robust Distances")
+    ylabel!("Squared Normalized Residuals")
+end
+
+end # end of module

src/dataimage.jl

@@ -1,14 +1,11 @@
 module DataImage
 
-
+export dataimage
 
 import ..Diagnostics:
     mahalanobisSquaredMatrix, euclideanDistances, mahalanobisSquaredBetweenPairs
 
-
-import Plots: RGB
-
-
+import ..RGB 
 
 """
 
@@ -41,11 +38,15 @@ julia> Plots.plot(di)
 # References
 Marchette, David J., and Jeffrey L. Solka. "Using data images for outlier detection." 
 Computational Statistics & Data Analysis 43.4 (2003): 541-552.
+
+!!! warning "Dependencies"
+    This method is enabled when the Plots package is installed and loaded.
+    
 """
 function dataimage(
     dataMatrix::Array{Float64,2};
     distance = :mahalanobis,
-)::Array{RGB{Float64},2}
+)::Matrix{RGB{Float64}}
     d = nothing
     if distance == :mahalanobis
         d = mahalanobisSquaredBetweenPairs(dataMatrix)

src/mveltsplot.jl

@@ -9,7 +9,6 @@ import ..MVE: mve
 
 import Distributions: Chisq, quantile
 
-using Plots
 
 """
     mveltsplot(setting; alpha = 0.05, showplot = true)
@@ -41,6 +40,10 @@ good leverage points (observations far from the remaining of data in both x and
 # References
 Van Aelst, Stefan, and Peter Rousseeuw. "Minimum volume ellipsoid." Wiley 
 Interdisciplinary Reviews: Computational Statistics 1.1 (2009): 71-82.
+
+!!! warning "Dependencies"
+    This method is enabled when the Plots package is installed and loaded.
+    
 """
 function mveltsplot(setting::RegressionSetting; alpha = 0.05, showplot = true)
     ltsresult = lts(setting)