Add user interface for automated classification (#222)

shapeout/gui/controls_analyze.py

@@ -14,6 +14,68 @@
 from . import confparms
 from .controls_subpanel import SubPanel
 
+
+
+class ClassifyStringDialog(wx.Dialog):
+
+    def __init__(self, analysis, parent, *args, **kw):
+        super(ClassifyStringDialog, self).__init__(parent, *args, **kw)
+        self.analysis = analysis
+        self.parent = parent
+        self.InitUI()
+        self.SetSize((250, 200))
+        self.SetTitle("Treatment and control identifiers")
+
+
+    def InitUI(self):
+        grid = wx.GridBagSizer(hgap=3)
+
+        grid.Add(wx.StaticText(self, label="Control"), (0,0))
+        grid.Add(wx.StaticText(self, label="Treatment"), (1,0))
+        grid.Add(wx.StaticText(self, label="Reservoir Control"), (2,0))
+        grid.Add(wx.StaticText(self, label="Reservoir Treatment"), (3,0))
+
+        self.wxctl = wx.TextCtrl(self, value="control")
+        self.wxtrt = wx.TextCtrl(self, value="")
+        self.wxctl_res = wx.TextCtrl(self, value="")
+        self.wxtrt_res = wx.TextCtrl(self, value="")
+
+        grid.Add(self.wxctl, (0,1))
+        grid.Add(self.wxtrt, (1,1))
+        grid.Add(self.wxctl_res, (2,1))
+        grid.Add(self.wxtrt_res, (3,1))
+
+        applyButton = wx.Button(self, label='Apply')
+        closeButton = wx.Button(self, label='Close')
+
+        grid.Add(applyButton, (4,0))
+        grid.Add(closeButton, (4,1))
+
+        self.SetSizer(grid)
+
+        applyButton.Bind(wx.EVT_BUTTON, self.OnApply)
+        closeButton.Bind(wx.EVT_BUTTON, self.OnClose)
+
+
+    def OnClose(self, e):
+        self.Destroy()
+
+
+    def OnApply(self, e):
+        treatment, repetition = lin_mix_mod.classify_treatment_repetition(
+            analysis=self.analysis,
+            id_ctl=self.wxctl.GetValue(),
+            id_trt=self.wxtrt.GetValue(),
+            id_ctl_res=self.wxctl_res.GetValue(),
+            id_trt_res=self.wxtrt_res.GetValue())
+
+        for mm, trt, rep in zip(self.analysis, treatment, repetition):
+            mm.config["analysis"]["regression treatment"] = trt
+            mm.config["analysis"]["regression repetition"] = rep
+
+        self.parent.update_classification()
+
+
 class SubPanelAnalyze(SubPanel):
     def __init__(self, parent, *args, **kwargs):
         SubPanel.__init__(self, parent, *args, **kwargs)
@@ -59,11 +121,16 @@ def make_analysis_choices(self, analysis):
             self.WXCB_axes.SetSelection(axid)
             sizer_bag.Add(self.WXCB_axes, (1,1), span=wx.GBSpan(1,2),
                           flag=wx.EXPAND|wx.ALL)
-
+
+
+            btn_auto = wx.Button(self, label="Autodetect classification")
+            sizer_bag.Add(btn_auto, (2,1), span=wx.GBSpan(1,2), flag=wx.EXPAND)
+            self.Bind(wx.EVT_BUTTON, self.OnAuto, btn_auto)
+
             # Header for table
-            sizer_bag.Add(wx.StaticText(self, label="Data set"), (2,0), span=wx.GBSpan(1,1))
-            sizer_bag.Add(wx.StaticText(self, label="Interpretation"), (2,1), span=wx.GBSpan(1,1))
-            sizer_bag.Add(wx.StaticText(self, label="Repetition"), (2,2), span=wx.GBSpan(1,1))
+            sizer_bag.Add(wx.StaticText(self, label="Data set"), (3,0), span=wx.GBSpan(1,1))
+            sizer_bag.Add(wx.StaticText(self, label="Interpretation"), (3,1), span=wx.GBSpan(1,1))
+            sizer_bag.Add(wx.StaticText(self, label="Repetition"), (3,2), span=wx.GBSpan(1,1))
 
             treatments = ["None", "Control", "Treatment",
                           "Reservoir Control", "Reservoir Treatment"]
@@ -73,7 +140,7 @@ def make_analysis_choices(self, analysis):
 
             for ii, mm in enumerate(analysis.measurements):
                 # title
-                sizer_bag.Add(wx.StaticText(self, label=mm.title), (3+ii,0), span=wx.GBSpan(1,1))
+                sizer_bag.Add(wx.StaticText(self, label=mm.title), (4+ii,0), span=wx.GBSpan(1,1))
                 # treatment
                 cbgtemp = wx.ComboBox(self, -1, choices=treatments,
                                       name=mm.identifier,
@@ -85,17 +152,14 @@ def make_analysis_choices(self, analysis):
                         mm.config["experiment"]["date"],
                         mm.config["experiment"]["time"])
                     cbgtemp.SetToolTip(wx.ToolTip(tip))
-                if mm.title.lower().count("control") or ii==0:
-                    cbgtemp.SetSelection(1)
-                else:
-                    cbgtemp.SetValue(mm.config["analysis"]["regression treatment"])
-                sizer_bag.Add(cbgtemp, (3+ii,1), flag=wx.EXPAND|wx.ALL)
+                cbgtemp.SetValue(mm.config["analysis"]["regression treatment"])
+                sizer_bag.Add(cbgtemp, (4+ii,1), flag=wx.EXPAND|wx.ALL)
                 # repetition
                 cbgtemp2 = wx.wx.SpinCtrl(self, -1, min=1, max=999,
                                           initial=1)
-                cbgtemp2.SetValue(mm.config["analysis"]["regression repetition"]-1)
+                cbgtemp2.SetValue(mm.config["analysis"]["regression repetition"])
 
-                sizer_bag.Add(cbgtemp2, (3+ii,2), flag=wx.EXPAND|wx.ALL)
+                sizer_bag.Add(cbgtemp2, (4+ii,2), flag=wx.EXPAND|wx.ALL)
 
                 self.WXCB_treatment.append(cbgtemp)
                 self.WXCB_repetition.append(cbgtemp2)
@@ -156,6 +220,11 @@ def OnApply(self, e=None):
         webbrowser.open(fd.name)
 
 
+    def OnAuto(self, e=None):
+        dlg = ClassifyStringDialog(self.analysis, self)
+        dlg.ShowModal()
+
+
     def OnReset(self, e=None):
         """
         Reset everything in the analysis tab.
@@ -193,6 +262,14 @@ def update_info_text(self, e=None):
         self.Layout()
 
 
+    def update_classification(self):
+        for mm, wxtrt, wxrep in zip(self.analysis,
+                                    self.WXCB_treatment,
+                                    self.WXCB_repetition):
+            wxtrt.SetValue(mm.config["analysis"]["regression treatment"])
+            wxrep.SetValue(mm.config["analysis"]["regression repetition"])
+
+
     def UpdatePanel(self, analysis=None):
         if analysis is None:
             analysis = self.analysis

shapeout/lin_mix_mod.py

@@ -59,8 +59,13 @@ def classify_treatment_repetition(analysis, id_ctl="co", id_trt="",
         else:
             idlist.append(["none", mm])
 
-    # extract treatment
-    treatment = [tt if tt != "none" else None for (tt, mm) in idlist]
+    # extract and rename treatment
+    treatment = [tt for (tt, mm) in idlist]
+    treatment = [tt.replace("res", "Reservoir") for tt in treatment]
+    treatment = [tt.replace("ctl", "Control") for tt in treatment]
+    treatment = [tt.replace("trt", "Treatment") for tt in treatment]
+    treatment = [tt.replace("none", "None") for tt in treatment]
+
     assert len(treatment) == len(analysis)
     # identify timeunit via similarity analysis
     ctl_str = [mm.title if tt == "ctl" else "" for (tt, mm) in idlist]

tests/test_linmixmod2.py

@@ -9,29 +9,32 @@
 """
 from __future__ import division, print_function, unicode_literals
 
+import copy
+
 import numpy as np
 
 from shapeout.lin_mix_mod import linmixmod, diffdef
 
 
-def test_linmixmod():
-    xs = [
-        [100, 99, 80, 120, 140, 150, 100, 100, 110, 111,
-            140, 145],  # Larger values (Control Channel1)
-        [20, 10, 5, 16, 14, 22, 27, 26, 5, 10, 11, 8, 15, 17,
-            20, 9],  # Smaller values (Control Reservoir1)
-        [115, 110, 90, 110, 145, 155, 110, 120, 115, 120, 120,
-            150, 100, 90, 100],  # Larger values (Control Channel2)
-        [30, 30, 15, 26, 24, 32, 37, 36, 15, 20, 21, 18, 25,
-            27, 30, 19],  # Smaller values (Control Reservoir2)
-        [150, 150, 130, 170, 190, 250, 150, 150, 160, 161, 180, 195, 130,
-            120, 125, 130, 125],  # Larger values (Treatment Channel1)
-        [2, 1, 5, 6, 4, 2, 7, 6, 5, 10, 1, 8, 5, 7, 2, 9, 11,
-            8, 13],  # Smaller values (Treatment Reservoir1)
-        [155, 155, 135, 175, 195, 255, 155, 155, 165, 165, 185, 200, 135, 125,
-            130, 135, 140, 150, 135, 140],  # Larger values (Treatment Channel2)
-        [25, 15, 19, 26, 44, 42, 35, 20, 15, 10, 11, 28, 35, 10, 25, 13]]  # Smaller values (Treatment Reservoir2)
+xs = [
+    [100, 99, 80, 120, 140, 150, 100, 100, 110, 111,
+        140, 145],  # Larger values (Control Channel1)
+    [20, 10, 5, 16, 14, 22, 27, 26, 5, 10, 11, 8, 15, 17,
+        20, 9],  # Smaller values (Control Reservoir1)
+    [115, 110, 90, 110, 145, 155, 110, 120, 115, 120, 120,
+        150, 100, 90, 100],  # Larger values (Control Channel2)
+    [30, 30, 15, 26, 24, 32, 37, 36, 15, 20, 21, 18, 25,
+        27, 30, 19],  # Smaller values (Control Reservoir2)
+    [150, 150, 130, 170, 190, 250, 150, 150, 160, 161, 180, 195, 130,
+        120, 125, 130, 125],  # Larger values (Treatment Channel1)
+    [2, 1, 5, 6, 4, 2, 7, 6, 5, 10, 1, 8, 5, 7, 2, 9, 11,
+        8, 13],  # Smaller values (Treatment Reservoir1)
+    [155, 155, 135, 175, 195, 255, 155, 155, 165, 165, 185, 200, 135, 125,
+        130, 135, 140, 150, 135, 140],  # Larger values (Treatment Channel2)
+    [25, 15, 19, 26, 44, 42, 35, 20, 15, 10, 11, 28, 35, 10, 25, 13]]  # Smaller values (Treatment Reservoir2)
+
 
+def test_linmixmod_1():
     # 1.: Differential Deformation in a linear mixed model
     treatment1 = ['Control', 'Reservoir Control', 'Control', 'Reservoir Control',
                   'Treatment', 'Reservoir Treatment', 'Treatment', 'Reservoir Treatment']
@@ -44,15 +47,22 @@ def test_linmixmod():
     assert np.allclose(
         [Result_1["p-Value (Likelihood Ratio Test)"]], [0.000602622503360039])
 
+
+def test_linmixmod_2():
     # 2.: Differential Deformation in a generalized linear mixed model
-    Result_2 = linmixmod(xs=xs, treatment=treatment1,
-                         timeunit=timeunit1, model='glmm')
+    treatment2 = ['Control', 'Reservoir Control', 'Control', 'Reservoir Control',
+                  'Treatment', 'Reservoir Treatment', 'Treatment', 'Reservoir Treatment']
+    timeunit2 = [1, 1, 2, 2, 1, 1, 2, 2]
+    Result_2 = linmixmod(xs=xs, treatment=treatment2,
+                         timeunit=timeunit2, model='glmm')
     assert np.allclose([Result_2["Estimate"]], [4.5385848573783898])
-    assert 'BOOTSTAP-DISTRIBUTIONS' in Result_1['Full Summary']
+    assert 'BOOTSTAP-DISTRIBUTIONS' in Result_2['Full Summary']
     assert 'GENERALIZED' in Result_2['Full Summary']
     assert np.allclose(
         [Result_2["p-Value (Likelihood Ratio Test)"]], [0.000556063024310929])
 
+
+def test_linmixmod_3():
     # 3.: Original values in a linear mixed model
     #'Reservoir' Measurements are now Controls and 'Channel' measurements are Treatments
     # This does not use differential deformation in linmixmod()
@@ -67,7 +77,9 @@ def test_linmixmod():
     assert np.allclose(
         [Result_3["p-Value (Likelihood Ratio Test)"]], [0.000331343267412872])
 
-    # 3.: Original values in a generalized linear mixed model
+
+def test_linmixmod_4():
+    # 4.: Original values in a generalized linear mixed model
     # This does not use differential deformation in linmixmod()
     treatment4 = ['Treatment', 'Control', 'Treatment', 'Control',
                   'Treatment', 'Control', 'Treatment', 'Control']

tests/test_linmixmod_classification.py

@@ -45,7 +45,7 @@ def test_classify_treatment_repetition_sinple():
         measurements += [ctl, trt]
     ana = Analysis(measurements)
     treatment, repetition = classify_treatment_repetition(ana, id_ctl="control")
-    assert treatment == ["ctl", "trt"] * 5
+    assert treatment == ["Control", "Treatment"] * 5
     assert np.all(repetition == np.repeat(np.arange(5), 2) + 1)
 
 
@@ -73,7 +73,10 @@ def test_classify_treatment_repetition_advanced():
     # Repetitions 1 and 2 are put in the end b/c "11" and "12" are better
     # matches than "1" and "2".
     repetition = np.roll(repetition, 8)
-    assert treatment == ["ctl", "trt", "res ctl", "res trt"] * 12
+    assert treatment == ["Control",
+                         "Treatment",
+                         "Reservoir Control",
+                         "Reservoir Treatment"] * 12
     assert np.all(repetition == np.repeat(np.arange(12), 4) + 1)