Miscellanous fixes (#956)

* Set no-ml default config properly for MultiClass

* Missing prefix for results all data

* Detect not found MC histos in fitter

* Do not load tensorflow et al. in apply() until we really want ML

* Fix pylint

machine_learning_hep/analysis/analyzerdhadrons.py

@@ -61,7 +61,7 @@ def __init__(self, datap, case, typean, period):
         self.d_resultsallpmc = self.d_prefix_mc + dp["mc"]["results"][period] \
             if period is not None \
             else self.d_prefix_mc + dp["mc"]["resultsallp"]
-        self.d_resultsallpdata =  + dp["data"]["results"][period] \
+        self.d_resultsallpdata = self.d_prefix_data + dp["data"]["results"][period] \
             if period is not None \
             else self.d_prefix_data + dp["data"]["resultsallp"]
 
@@ -330,17 +330,20 @@ def makenormyields(self):  # pylint: disable=import-outside-toplevel, too-many-b
             hbkg.Write()
             fileoutbkgscaled.Close()
 
+        output_prompt = []
         hf_pt_spectrum(self.p_anahpt,
                            self.p_br,
                            self.p_inputfonllpred,
                            self.p_fd_method,
+                           None,
                            fileouteff,
                            namehistoeffprompt,
                            namehistoefffeed,
                            yield_filename,
                            nameyield,
                            selnorm,
                            self.p_sigmamb,
+                           output_prompt,
                            fileoutcross)
 
         fileoutcrosstot = TFile.Open("%s/finalcross%s%stot.root" %

machine_learning_hep/config.py

@@ -62,8 +62,13 @@ def update_config(database: dict, run_config: dict, database_overwrite=None): #
             database["mlapplication"][keys[0]][keys[1]][:] = \
                     [f"{path}_std" for path in database["mlapplication"][keys[0]][keys[1]]]
         # ...set the ML working point all to 0
+        # except for MultiClassification, where bkg cut of 1 is the loosest one
         for k in data_mc:
-            database["mlapplication"]["probcutpresel"][k][:] = \
-                    [0] * len(database["mlapplication"]["probcutpresel"][k])
-        database["mlapplication"]["probcutoptimal"][:] \
-                = [0] * len(database["mlapplication"]["probcutoptimal"])
+            database["mlapplication"]["probcutpresel"][k] = \
+              [[1 if i == 0 and database["ml"]["mltype"] == "MultiClassification" else 0 \
+                for i in range(len(pcut))] \
+                for pcut in database["mlapplication"]["probcutpresel"][k]]
+        database["mlapplication"]["probcutoptimal"] = \
+          [[1 if i == 0 and database["ml"]["mltype"] == "MultiClassification" else 0 \
+            for i in range(len(pcut))] \
+            for pcut in database["mlapplication"]["probcutoptimal"]]

machine_learning_hep/fitting/helpers.py

@@ -347,9 +347,15 @@ def get_histograms(self, ibin1, ibin2, get_data=True, get_mc=False, get_reflecti
         histo_reflections = None
         if get_mc:
             histo_mc = file_mc.Get("hmass_sig" + suffix)
+            if not histo_mc:
+                file_mc.ls()
+                raise ValueError("Did not find", "hmass_sig" + suffix)
             histo_mc.SetDirectory(0)
         if get_reflections:
             histo_reflections = file_mc.Get("hmass_refl" + suffix)
+            if not histo_reflections:
+                file_mc.ls()
+                raise ValueError("Did not find", "hmass_refl" + suffix)
             histo_reflections.SetDirectory(0)
         file_mc.Close()
 

machine_learning_hep/processer.py

@@ -457,8 +457,6 @@ def skim(self, file_index):
                 write_df(dfgensk, self.mptfiles_gensk[ipt][file_index])
 
     def applymodel(self, file_index):
-        from machine_learning_hep.models import \
-            apply  # pylint: disable=import-error, import-outside-toplevel
         for ipt in range(self.p_nptbins):
             if os.path.exists(self.mptfiles_recoskmldec[ipt][file_index]):
                 if os.stat(self.mptfiles_recoskmldec[ipt][file_index]).st_size != 0:
@@ -467,6 +465,8 @@ def applymodel(self, file_index):
             if self.p_mask_values:
                 mask_df(dfrecosk, self.p_mask_values)
             if self.doml is True:
+                from machine_learning_hep.models import \
+                    apply  # pylint: disable=import-error, import-outside-toplevel
                 if os.path.isfile(self.lpt_model[ipt]) is False:
                     print("Model file not present in bin %d" % ipt)
                 with openfile(self.lpt_model[ipt], 'rb') as mod_file: