bug fixes

data/solverLog.csv

@@ -125,3 +125,45 @@ Uni-criterion: clusterCenters,greedySwap,1000,2,10,0.13956262497231364,0.3799413
 "Uni-criterion: preserveMetric, hull",greedyMinSubset,500,2,12,0.2948356249835342,0.0
 "Uni-criterion: sum, outlierness",greedySwap,500,2,40,0.07649645791389048,-74.23043000890738
 "Uni-criterion: distinctness, distances",greedySwap,500,2,60,0.5947001250460744,-92.08564918686247
+"Uni-criterion: preserveMetric, hull",greedyMinSubset,500,2,9,0.20904829213395715,0.0
+"Uni-criterion: sum, outlierness",greedySwap,500,2,40,0.08821466704830527,-79.95507869120974
+"Uni-criterion: distinctness, distances",greedySwap,500,2,60,0.6470217080786824,-94.69010039466121
+"Uni-criterion: preserveMetric, hull",greedyMinSubset,500,2,10,0.221585167106241,0.0
+"Uni-criterion: sum, outlierness",greedySwap,500,2,40,0.08099545910954475,-76.8424909434427
+"Uni-criterion: distinctness, distances",greedySwap,500,2,60,0.6626855423673987,-87.75324874515424
+"Uni-criterion: preserveMetric, hull",greedyMinSubset,200,2,9,0.07839837530627847,0.0
+"Uni-criterion: sum, outlierness",greedySwap,200,2,40,0.03452433319762349,-76.73330214179785
+"Uni-criterion: distinctness, distances",greedySwap,200,2,60,0.15389200020581484,-72.30676106510778
+"Multi-criterion: 100*(earthMoversDistance) + 1*(distinctness, distances)",greedySwap,200,2,80,8.683466792106628,-37.832634391613226
+"Multi-criterion: 50*(earthMoversDistance) + 1*(distinctness, distances)",greedySwap,200,2,80,8.944492833223194,-55.336713455218515
+"Multi-criterion: 1*(earthMoversDistance) + 1*(distinctness, distances)",greedySwap,200,2,80,10.16267108405009,-77.45551739898436
+"Uni-criterion: preserveMetric, hull",greedyMinSubset,200,2,9,0.07779270829632878,0.0
+"Uni-criterion: sum, outlierness",greedySwap,200,2,40,0.029774457681924105,-70.13661593006383
+"Uni-criterion: distinctness, distances",greedySwap,200,2,60,0.1661411253735423,-76.52681602089089
+"Multi-criterion: 100*(earthMoversDistance) + 1*(distinctness, distances)",greedySwap,200,2,80,8.791529291309416,-35.549221713913795
+"Multi-criterion: 50*(earthMoversDistance) + 1*(distinctness, distances)",greedySwap,200,2,80,9.541060833260417,-56.63750965002805
+"Multi-criterion: 1*(earthMoversDistance) + 1*(distinctness, distances)",greedySwap,200,2,80,10.46921870810911,-83.31100611009497
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,12,0.0214225840754807,-37.26355923400367
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,36,0.0722177498973906,-57.328890897044175
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,28,0.05338112497702241,-58.87100686028796
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,17,0.03293449990451336,-44.09659907116473
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,55,0.11244387505576015,-61.49184399076313
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,101,0.24100016662850976,-86.10404158691499
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,15,0.031042208895087242,-37.86319298068655
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,50,0.10623954189941287,-61.01378931117458
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,103,0.24973824992775917,-83.00438182523874
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,20,0.0373823749832809,-45.76776906728566
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,11,0.017222166061401367,-44.268067171984875
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,92,0.2180039999075234,-82.50779303999899
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,23,0.04364325013011694,-44.964197525236315
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,52,0.10621762461960316,-61.43036396431012
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,85,0.2009903327561915,-79.26661008322394
+"Uni-criterion: preserveMetric, hull",greedyMinSubset,200,2,9,0.06682970793917775,0.0
+"Uni-criterion: sum, outlierness",greedySwap,200,2,40,0.03168637491762638,-87.25866709672482
+"Uni-criterion: distinctness, distances",greedySwap,200,2,60,0.17151008266955614,-83.69038566035452
+"Multi-criterion: 100*(earthMoversDistance) + 1*(distinctness, distances)",greedySwap,200,2,80,9.204857917036861,-27.184808100622732
+"Multi-criterion: 10*(earthMoversDistance) + 1*(distinctness, distances)",greedySwap,200,2,80,10.099580999929458,-78.56242065843928
+"Multi-criterion: 1*(earthMoversDistance) + 1*(distinctness, distances)",greedySwap,200,2,80,10.901172333396971,-87.1982622784361
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,21,0.04174462519586086,-47.546951611341385
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,13,0.021749832667410374,-51.19458441796628
+"Uni-criterion: distinctness, distances",greedyMixed,200,2,108,0.24733683420345187,-88.67659171843289

flexibleSubsetSelection/plot.py

@@ -156,7 +156,7 @@ def errorMarkers(ax, x, vals1, color1, marker1, vals2=None, color2=None,
 
 # --- Plots --------------------------------------------------------------------
 
-def initialize(color, font="Times New Roman", family="sans-serif", size=42):
+def initialize(color, font="Times New Roman", size=42):
     """
     Initialize matplotlib settings global parameters for text and background
     
@@ -171,8 +171,7 @@ def initialize(color, font="Times New Roman", family="sans-serif", size=42):
     if not isinstance(color, Color):
         raise ValueError("color must be an instance of Color object")
 
-    plt.rcParams["font.sans-serif"] = font
-    plt.rcParams["font.family"] = family
+    plt.rcParams["font.family"] = font
     plt.rcParams["pdf.fonttype"] = size
     plt.rcParams["ps.fonttype"] = size
     plt.rcParams["axes.facecolor"] = color.palette["grey"]

jupyter/Fig1-designProcess.ipynb

@@ -33,7 +33,7 @@
     "# Initialize notebook settings\n",
     "sns.set_theme() # set seaborn theme\n",
     "matplotlib_inline.backend_inline.set_matplotlib_formats('svg') # vector plots\n",
-    "%matplotlib inline \n",
+    "# %matplotlib inline \n",
     "%load_ext autoreload\n",
     "%autoreload 2"
    ]
@@ -58,7 +58,7 @@
     "seed = 123456789                 # random generation seed for replicability\n",
     "\n",
     "# Create a random blobs dataset to use as our example dataset\n",
-    "dataset = fss.Dataset(randTypes=\"blobs\", size=(500, 2), seed=seed)\n",
+    "dataset = fss.Dataset(randTypes=\"blobs\", size=(200, 2), seed=seed)\n",
     "dataset.save(f\"{directory}/fullData\")"
    ]
   },
@@ -104,8 +104,12 @@
     "\n",
     "Applying this objective allows us to select a subset of 40 points with the highest local outlier effect.\n",
     "\n",
-    "$$\\max_{S \\in \\mathbb{S}} \\space \\text{LOF}(S)$$\n",
-    "$$\\text{s.t.} \\space \\|S\\| = 40$$"
+    "$$\n",
+    "\\max_{S \\in \\mathbb{S}} \\ \\text{LOF}(S)\n",
+    "$$\n",
+    "$$\n",
+    "\\text{s.t.} \\ \\|S\\| = 40\n",
+    "$$"
    ]
   },
   {
@@ -173,23 +177,23 @@
    "source": [
     "# Initialize color and plot settings\n",
     "color = fss.plot.Color()\n",
-    "fss.plot.initialize(color)\n",
+    "fss.plot.initialize(color, font=\"DejaVu Sans\")\n",
     "\n",
     "# Plot the three different resulting subsets as scatterplots\n",
     "titleSize = 24\n",
     "subtitleSize = 18\n",
     "titles = [\"Hull\", \"Outliers\", \"Distinctness\"]\n",
     "subsets = [subsetHull, subsetOutliers, subsetDistinct]\n",
     "\n",
-    "fig, axs = plt.subplots(nrows=1, ncols=3, figsize=(6.25, 3))\n",
+    "fig, axs = plt.subplots(nrows=1, ncols=3, figsize=(8, 3.5))\n",
     "fig.text(0.49, 1, '1. Express', ha='center', va='center', fontsize=titleSize)\n",
     "\n",
     "for i, ax in enumerate(fig.axes):\n",
     "    ax.grid(visible=False)\n",
     "    ax.set_xticks([])\n",
     "    ax.set_yticks([])\n",
     "    ax.set_title(titles[i], fontsize=subtitleSize)\n",
-    "    ax.set_aspect(\"equal\")\n",
+    "    # ax.set_aspect(\"equal\")\n",
     "\n",
     "    fss.plot.scatter(ax = ax, \n",
     "                 color = color, \n",
@@ -218,23 +222,23 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# subsetSize = 80 # The size of subsets being selected with blended objectives\n",
+    "subsetSize = 80 # The size of subsets being selected with blended objectives\n",
     "\n",
-    "# # Use distribution and distinctness objectives\n",
-    "# objectives = [fss.objective.earthMoversDistance, fss.objective.distinctness]\n",
+    "# Use distribution and distinctness objectives\n",
+    "objectives = [fss.objective.earthMoversDistance, fss.objective.distinctness]\n",
     "\n",
-    "# # Parameters of the distribution and distinctness objectives\n",
-    "# parameters = [{\"dataset\": dataset.dataArray}, \n",
-    "#               {\"solveArray\": \"distances\", \"selectBy\": \"matrix\"}]\n",
+    "# Parameters of the distribution and distinctness objectives\n",
+    "parameters = [{\"dataset\": dataset.dataArray}, \n",
+    "              {\"solveArray\": \"distances\", \"selectBy\": \"matrix\"}]\n",
     "\n",
-    "# # Create the multicriterion loss function from the objectives and weight them\n",
-    "# solver.lossFunction = fss.MultiCriterion(objectives = objectives, \n",
-    "#                                          parameters = parameters, \n",
-    "#                                          weights=[100, 1])\n",
+    "# Create the multicriterion loss function from the objectives and weight them\n",
+    "solver.lossFunction = fss.MultiCriterion(objectives = objectives, \n",
+    "                                         parameters = parameters, \n",
+    "                                         weights=[100, 1])\n",
     "\n",
-    "# # Solve for the blended distribution and distinctness subset\n",
-    "# subsetBlend1 = solver.solve(dataset, subsetSize=subsetSize, verbose=verbose)\n",
-    "# subsetBlend1.save(f\"{directory}/blend1Subset\")"
+    "# Solve for the blended distribution and distinctness subset\n",
+    "subsetBlend1 = solver.solve(dataset, subsetSize=subsetSize, verbose=verbose)\n",
+    "subsetBlend1.save(f\"{directory}/blend1Subset\")"
    ]
   },
   {
@@ -243,14 +247,14 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# # Update the weights to provide less emphasis on the distribution objective\n",
-    "# solver.lossFunction = fss.MultiCriterion(objectives = objectives, \n",
-    "#                                          parameters = parameters, \n",
-    "#                                          weights=[50, 1])\n",
-    "\n",
-    "# # Solve for the blended distribution and distinctness subset\n",
-    "# subsetBlend2 = solver.solve(dataset, subsetSize=subsetSize, verbose=verbose)\n",
-    "# subsetBlend2.save(f\"{directory}/blend2Subset\")"
+    "# Update the weights to provide less emphasis on the distribution objective\n",
+    "solver.lossFunction = fss.MultiCriterion(objectives = objectives, \n",
+    "                                         parameters = parameters, \n",
+    "                                         weights=[10, 1])\n",
+    "\n",
+    "# Solve for the blended distribution and distinctness subset\n",
+    "subsetBlend2 = solver.solve(dataset, subsetSize=subsetSize, verbose=verbose)\n",
+    "subsetBlend2.save(f\"{directory}/blend2Subset\")"
    ]
   },
   {
@@ -259,14 +263,14 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# # Update the weights to an even weight of the two objectives\n",
-    "# solver.lossFunction = fss.MultiCriterion(objectives = objectives, \n",
-    "#                                          parameters = parameters, \n",
-    "#                                          weights=[1, 1])\n",
-    "\n",
-    "# # Solve for the blended distribution and distinctness subset\n",
-    "# subsetBlend3 = solver.solve(dataset, subsetSize=subsetSize, verbose=verbose)\n",
-    "# subsetBlend3.save(f\"{directory}/blend3Subset\")"
+    "# Update the weights to an even weight of the two objectives\n",
+    "solver.lossFunction = fss.MultiCriterion(objectives = objectives, \n",
+    "                                         parameters = parameters, \n",
+    "                                         weights=[1, 1])\n",
+    "\n",
+    "# Solve for the blended distribution and distinctness subset\n",
+    "subsetBlend3 = solver.solve(dataset, subsetSize=subsetSize, verbose=verbose)\n",
+    "subsetBlend3.save(f\"{directory}/blend3Subset\")"
    ]
   },
   {
@@ -284,28 +288,28 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# # Plot the three subsets with different blends of the two objectives\n",
-    "# titles = [\"More Distribution,\\nLess Distinct\", \"\", \n",
-    "#           \"Less Distribution,\\nMore Distinct\"]\n",
-    "# subsets = [subsetBlend1, subsetBlend2, subsetBlend3]\n",
-    "\n",
-    "# fig, axs = plt.subplots(nrows=1, ncols=3, figsize=(6.25, 3))\n",
-    "# fig.text(0.49, 1, '2. Blend', ha='center', va='center', fontsize=titleSize)\n",
-    "\n",
-    "# for i, ax in enumerate(fig.axes):\n",
-    "#     ax.grid(visible=False)\n",
-    "#     ax.set_xticks([])\n",
-    "#     ax.set_yticks([])\n",
-    "#     ax.set_title(titles[i], fontsize=subtitleSize)\n",
-    "#     ax.set_aspect(\"equal\")\n",
-    "\n",
-    "#     fss.plot.scatter(ax = ax, \n",
-    "#                  color = color, \n",
-    "#                  dataset = dataset, \n",
-    "#                  subset = subsets[i], \n",
-    "#                  alpha = 0.6)\n",
-    "\n",
-    "# plt.savefig(f\"../figures/{directory}/blend.pdf\", bbox_inches=\"tight\")"
+    "# Plot the three subsets with different blends of the two objectives\n",
+    "titles = [\"More Distribution,\\nLess Distinct\", \"\", \n",
+    "          \"Less Distribution,\\nMore Distinct\"]\n",
+    "subsets = [subsetBlend1, subsetBlend2, subsetBlend3]\n",
+    "\n",
+    "fig, axs = plt.subplots(nrows=1, ncols=3, figsize=(8, 3.5))\n",
+    "fig.text(0.49, 1, '2. Blend', ha='center', va='center', fontsize=titleSize)\n",
+    "\n",
+    "for i, ax in enumerate(fig.axes):\n",
+    "    ax.grid(visible=False)\n",
+    "    ax.set_xticks([])\n",
+    "    ax.set_yticks([])\n",
+    "    ax.set_title(titles[i], fontsize=subtitleSize)\n",
+    "    ax.set_aspect(\"equal\")\n",
+    "\n",
+    "    fss.plot.scatter(ax = ax, \n",
+    "                 color = color, \n",
+    "                 dataset = dataset, \n",
+    "                 subset = subsets[i], \n",
+    "                 alpha = 0.6)\n",
+    "\n",
+    "plt.savefig(f\"../figures/{directory}/blend.pdf\", bbox_inches=\"tight\")"
    ]
   },
   {
@@ -345,7 +349,7 @@
     "subsetDistinct2.save(f\"{directory}/distinct2Subset\")\n",
     "\n",
     "subsetDistinct3 = solver.solve(dataset = dataset, \n",
-    "                               weight = 0.1, \n",
+    "                               weight = 0.05, \n",
     "                               initialSize = 3, \n",
     "                               verbose = verbose)\n",
     "subsetDistinct3.save(f\"{directory}/distinct3Subset\")"
@@ -370,7 +374,7 @@
     "titles = [\"Fewer Points,\\nMore Distinct\", \"\", \"More Points,\\nLess Distinct\"]\n",
     "subsets = [subsetDistinct1, subsetDistinct2, subsetDistinct3]\n",
     "\n",
-    "fig, axs = plt.subplots(nrows=1, ncols=3, figsize=(6.25, 3))\n",
+    "fig, axs = plt.subplots(nrows=1, ncols=3, figsize=(8, 3.5))\n",
     "fig.text(0.49, 1, '3. Tune', ha='center', va='center', fontsize=titleSize)\n",
     "\n",
     "for i, ax in enumerate(fig.axes):\n",