update plots

replication/partition_plots/partition_plots.jl

@@ -19,10 +19,10 @@ end
 function plot_mondrian_tree(tree::MondrianTree)
     @assert isa(tree, MondrianTree{2})
     splits = get_splits(tree)
-    (fig, ax) = plt.subplots(figsize=(2.1, 2.1))
+    (fig, ax) = plt.subplots(figsize=(3, 3))
 
     # plot root cell
-    lw = 0.3
+    lw = 0.5
     (l1, l2) = tree.lower
     (u1, u2) = tree.upper
     plot([l1, l1], [l1, u2], color="k", lw=lw)
@@ -38,10 +38,10 @@ function plot_mondrian_tree(tree::MondrianTree)
     end
 
     # format plot
-    plt.xticks([0, 1])
-    plt.yticks([0, 1])
-    plt.xlabel("\$x_1\$")
-    plt.ylabel("\$x_2\$")
+    plt.xticks([0, 1], fontsize=11)
+    plt.yticks([0, 1], fontsize=11)
+    plt.xlabel("\$x_1\$", fontsize=12)
+    plt.ylabel("\$x_2\$", fontsize=12)
     ax.xaxis.set_label_coords(0.5, -0.04)
     ax.yaxis.set_label_coords(-0.04, 0.5)
     ax.tick_params(color="w", direction="in", pad=0)
@@ -61,7 +61,7 @@ for i in 1:length(lambdas)
     global lambda = lambdas[i]
     global tree = MondrianTree(d, lambda)
     global (fig, ax) = plot_mondrian_tree(tree)
-    savefig("./replication/partition_plots/plot_mondrian_process_$i.pgf", bbox_inches="tight")
-    savefig("./replication/partition_plots/plot_mondrian_process_$i.pdf", bbox_inches="tight")
+    plt.tight_layout()
+    savefig("./replication/partition_plots/plot_mondrian_process_$i.pdf")
     plt.close("all")
 end

replication/weather/weather.jl

@@ -48,16 +48,20 @@ function format_plot(ax)
     yticks = [990 + i * 10 for i in 0:5]
     xticklabels = "\$" .* string.(xticks) .* "\$"
     yticklabels = "\$" .* string.(yticks) .* "\$"
-    plt.xticks((xticks .- x_min) ./ (x_max - x_min), labels=xticklabels)
-    plt.yticks((yticks .- y_min) ./ (y_max - y_min), labels=yticklabels)
-    plt.xlabel("Relative humidity at 3pm (\\%)")
-    plt.ylabel("Pressure at 3pm (mbar)")
+    plt.xticks((xticks .- x_min) ./ (x_max - x_min), labels=xticklabels, fontsize=11)
+    plt.yticks((yticks .- y_min) ./ (y_max - y_min), labels=yticklabels, fontsize=11)
+    plt.xlabel("Relative humidity at 3pm (\\%)", fontsize=12)
+    plt.ylabel("Pressure at 3pm (mbar)", fontsize=12)
     # color key
+    #handle = plt.scatter([], [], c="white")
     dry_handle = plt.scatter([], [], c=dry_color)
     wet_handle = plt.scatter([], [], c=wet_color)
-    ax.legend([dry_handle, wet_handle], ["Dry tomorrow", "Wet tomorrow"],
+    ax.legend([wet_handle, dry_handle],
+              ["Rain next day", "No rain next day"],
               handletextpad=0.1, frameon=false,
-              bbox_to_anchor=(1.01, 1.16), ncol=2)
+              bbox_to_anchor=(1.04, 1.13), ncol=3,
+              fontsize=12,
+              columnspacing=0.8)
     # layout
     plt.xlim([-0.05, 1.05])
     plt.ylim([-0.05, 1.05])
@@ -221,7 +225,7 @@ limit = nothing
 (data, x_min, x_max, y_min, y_max) = load_data(limit=limit)
 dry_color = "#da6200"
 wet_color = "#0080d0"
-figsize = (3.5, 3.7)
+figsize = (4, 4.2)
 dpi = 500
 
 # make trees

replication/weather/weather_cv.jl

@@ -100,7 +100,7 @@ for lambda in lambdas
     println("mse: ", mse)
 end
 
-(fig, ax) = plt.subplots(figsize=(3.5, 3.7))
+(fig, ax) = plt.subplots(figsize=(4, 4.4))
 best_lambda = 5.0
 i = [i for i in 1:length(lambdas) if isapprox(lambdas[i], best_lambda, rtol=0.01)][]
 plt.plot([best_lambda, best_lambda], [0.0, gcvs[i] - 0.0001], c="#666677",
@@ -110,15 +110,15 @@ plt.plot(lambdas, mses, lw=1.0, c="#aa44dd",
 plt.plot(lambdas, gcvs, lw=1.0, c="#009944",
          label="Generalized cross-validation")
 plt.ylim([0.11 - 0.002, 0.17 + 0.002])
-plt.yticks(range(0.11, stop=0.17, step=0.01))
-plt.xlabel("Lifetime parameter \$\\lambda\$")
-plt.ylabel("Loss function")
-plt.legend(frameon=false)
-plt.subplots_adjust(left=0.205, right=0.96, top=0.854, bottom=0.165)
+plt.yticks(range(0.11, stop=0.17, step=0.01), fontsize=11)
+plt.xticks(fontsize=11)
+plt.xlabel("Lifetime parameter \$\\lambda\$", fontsize=12)
+plt.ylabel("Loss function", fontsize=12)
+plt.legend(fontsize=12)
+plt.subplots_adjust(left=0.205, right=0.96, top=0.842, bottom=0.140)
 plt.savefig("./replication/weather/weather_gcv.png", dpi=500)
 
 # CIs
-#limit = 1000
 limit = nothing
 (data, x_min, x_max, y_min, y_max) = load_data(limit=limit)
 n = nrow(data)