fix: correct display of KDE for scatter plot in log scale; ref: give …

…up making log-scale contours work with chaco (#231)

CHANGELOG

@@ -5,8 +5,7 @@
    by chaco testing whether an array object is contained in another
    object
  - Bugfixes:
-   - Workaround for plotting contours on a log scale using
-     https://github.com/enthought/chaco/issues/300
+   - Scatter plot events are now labeled correctly when log-scale is used
 0.9.1
  - Add tooltips for box filter settings
  - Rename *Statistical analysis* to *Statistical summary* in batch menu

shapeout/analysis.py

@@ -588,6 +588,13 @@ def SetParameters(self, newcfg):
                     pops.append(skey)
             for skey in pops:
                 pl.pop(skey)
+            # Address issue with faulty contour plot on log scale
+            # https://github.com/enthought/chaco/issues/300
+            if (("scale x" in pl and pl["scale x"] == "log") or
+                    ("scale y" in pl and pl["scale y"] == "log")):
+                warnings.warn(
+                    "Disabling contour plot because of chaco issue #300!")
+                pl["contour plot"] = False
             # check for inverted plotting ranges
             for feat in dfn.scalar_feature_names:
                 fmin = feat + " min"

shapeout/gui/plot_contour.py

@@ -113,6 +113,12 @@ def set_contour_data(plot, analysis):
     xax = mm.config["plotting"]["axis x"].lower()
     yax = mm.config["plotting"]["axis y"].lower()
 
+    scalex = mm.config["plotting"]["scale x"].lower()
+    scaley = mm.config["plotting"]["scale y"].lower()
+
+    plot.index_scale = scalex
+    plot.value_scale = scaley
+
     if mm.config["filtering"]["enable filters"]:
         x0 = mm[xax][mm._filter]
         y0 = mm[yax][mm._filter]
@@ -131,18 +137,29 @@ def set_contour_data(plot, analysis):
             break
 
         kde_type = mm.config["plotting"]["kde"].lower()
-        kde_kwargs = plot_common.get_kde_kwargs(x=x0, y=y0, kde_type=kde_type,
-                                                xacc=mm.config["plotting"]["kde accuracy "+xax],
-                                                yacc=mm.config["plotting"]["kde accuracy "+yax])
+        kde_kwargs = plot_common.get_kde_kwargs(
+            x=x0,
+            y=y0,
+            kde_type=kde_type,
+            xacc=mm.config["plotting"]["kde accuracy "+xax],
+            yacc=mm.config["plotting"]["kde accuracy "+yax])
         # Accuracy for plotting contour data
         xacc = mm.config["plotting"]["contour accuracy "+xax]
         yacc = mm.config["plotting"]["contour accuracy "+yax]
 
         a = time.time()
-        (X,Y,density) = mm.get_kde_contour(xax=xax, yax=yax, xacc=xacc, yacc=yacc,
-                                           kde_type=kde_type, kde_kwargs=kde_kwargs)
+        X, Y, density = mm.get_kde_contour(xax=xax,
+                                           yax=yax,
+                                           xacc=xacc,
+                                           yacc=yacc,
+                                           xscale=scalex,
+                                           yscale=scaley,
+                                           kde_type=kde_type,
+                                           kde_kwargs=kde_kwargs,
+                                           )
+
         print("...KDE contour time {}: {:.2f}s".format(kde_type, time.time()-a))
-        
+
         pd.set_data(cname, density)
 
         # contour widths
@@ -157,10 +174,15 @@ def set_contour_data(plot, analysis):
         if mode == "fraction":
             plev = list(np.nanmax(density) * levels)
         elif mode == "quantile":
-            pdensity = mm.get_kde_scatter(xax=xax, yax=yax,
+            pdensity = mm.get_kde_scatter(xax=xax,
+                                          yax=yax,
+                                          xscale=scalex,
+                                          yscale=scaley,
                                           kde_type=kde_type,
-                                          kde_kwargs=kde_kwargs)
+                                          kde_kwargs=kde_kwargs,
+                                          )
             plev = list(np.nanpercentile(pdensity, q=levels*100))
+
         else:
             raise ValueError("Unknown contour level mode `{}`!".format(mode))
 
@@ -171,23 +193,22 @@ def set_contour_data(plot, analysis):
             styles = "solid"
             widths = cwidth
 
-        scalex = mm.config["plotting"]["scale x"].lower()
-        scaley = mm.config["plotting"]["scale y"].lower()
-        plot.index_scale = scalex
-        plot.value_scale = scaley
-
         cplot = plot.contour_plot(cname,
                           name=cname,
                           type="line",
-                          xbounds=(X[0][0], X[0][-1]),
-                          ybounds=(Y[0][0], Y[-1][0]),
+                          xbounds=X,
+                          ybounds=Y,
                           levels=plev,
                           colors=mm.config["plotting"]["contour color"],
                           styles=styles,
                           widths=widths,
+                          index_scale=scalex,
+                          value_scale=scaley,
                           )[0]
         # Workaround for plotting contour data on a log scale
         # (https://github.com/enthought/chaco/issues/300)
+        # 2019-04-09: This does not resolve the problem. In case of a
+        # logarithmic scale, there is an offset in the contour plotted.
         if scalex == "log":
             cplot.index_mapper._xmapper = ca.LogMapper(
                     range=cplot.index_range.x_range,
@@ -196,5 +217,5 @@ def set_contour_data(plot, analysis):
         if scaley == "log":
             cplot.index_mapper._ymapper = ca.LogMapper(
                     range=cplot.index_range.y_range,
-                    screen_bounds=cplot.index_mapper.screen_bounds[2:]
+                    screen_bounds=cplot.index_mapper.screen_bounds[:2]
                 )

shapeout/gui/plot_scatter.py

@@ -208,13 +208,23 @@ def set_scatter_data(plot, mm):
 
 
     kde_type = mm.config["plotting"]["kde"].lower()
-    kde_kwargs = plot_common.get_kde_kwargs(x=x, y=y, kde_type=kde_type,
+
+    kde_kwargs = plot_common.get_kde_kwargs(x=x,y=y, kde_type=kde_type,
                                             xacc=mm.config["plotting"]["kde accuracy "+xax],
                                             yacc=mm.config["plotting"]["kde accuracy "+yax])
 
     a = time.time()
-    density = mm.get_kde_scatter(xax=xax, yax=yax, positions=positions,
-                                 kde_type=kde_type, kde_kwargs=kde_kwargs)
+    scalex = mm.config["plotting"]["scale x"].lower()
+    scaley = mm.config["plotting"]["scale y"].lower()
+
+    density = mm.get_kde_scatter(xax=xax,
+                                 yax=yax,
+                                 positions=positions,
+                                 kde_type=kde_type,
+                                 xscale=scalex,
+                                 yscale=scaley,
+                                 kde_kwargs=kde_kwargs,
+                                 )
     print("...KDE scatter time {}: {:.2f}s".format(kde_type, time.time()-a))
 
     pd = plot.data