bf

rkansal47 · Oct 26, 2023 · 32c5ad3 · 32c5ad3
1 parent 6c497b6
commit 32c5ad3
Show file tree

Hide file tree

Showing 4 changed files with 50 additions and 3 deletions.
diff --git a/README.md b/README.md
@@ -294,7 +294,7 @@ nohup bash -c 'for sample in NMSSM_XToYHTo2W2BTo4Q2B_MX-4000_MY-150 NMSSM_XToYHT
 
 ### Create Datacard
 
-Need `root==6.22.6`, and `square_coef` branch of https://github.com/rkansal47/rhalphalib installed (`pip install -e . --user` after checking out the branch). `CMSSW_11_2_0` recommended.
+Need `root==6.22.6`, and `square_coef` branch of https://github.com/rkansal47/rhalphalib installed (`pip install -e . --user` after checking out the branch).
 
 ```bash
 python3 postprocessing/CreateDatacard.py --templates-dir templates/$TAG --model-name $TAG (--resonant)

diff --git a/src/HHbbVV/postprocessing/CreateDatacard.py b/src/HHbbVV/postprocessing/CreateDatacard.py
@@ -750,7 +750,7 @@ def nonres_alphabet_fit(
         # was integer, and numpy complained about subtracting float from it
         initial_qcd = failCh.getObservation().astype(float)
         for sample in failCh:
-            if args.resonant and sample.sampletype == rl.Sample.SIGNAL:
+            if sample.sampletype == rl.Sample.SIGNAL:
                 continue
             logging.debug("subtracting %s from qcd" % sample._name)
             initial_qcd -= sample.getExpectation(nominal=True)

diff --git a/src/HHbbVV/postprocessing/PostProcess.ipynb b/src/HHbbVV/postprocessing/PostProcess.ipynb
@@ -77,7 +77,7 @@
     "# bdt_data_dir = \"/eos/uscms/store/user/cmantill/bbVV/skimmer/Jun10/bdt_data/\"\n",
     "year = \"2018\"\n",
     "\n",
-    "date = \"23Aug21\"\n",
+    "date = \"23Oct12\"\n",
     "plot_dir = f\"../../../plots/PostProcessing/{date}/\"\n",
     "templates_dir = f\"templates/{date}\"\n",
     "_ = os.system(f\"mkdir -p {plot_dir}\")\n",
@@ -173,6 +173,40 @@
     "postprocessing.load_bdt_preds(events_dict, year, bdt_preds_dir, jec_jmsr_shifts=True)"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Aside: checking if regressed mass is actually saving useful jets"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "bbmsd = utils.get_feat(events_dict[\"HHbbVV\"], \"bbFatJetMsd\", bb_masks[\"HHbbVV\"])\n",
+    "vvmsd = utils.get_feat(events_dict[\"HHbbVV\"], \"VVFatJetMsd\", bb_masks[\"HHbbVV\"])\n",
+    "bbpnetm = utils.get_feat(events_dict[\"HHbbVV\"], \"bbFatJetParticleNetMass\", bb_masks[\"HHbbVV\"])\n",
+    "vvpnetm = utils.get_feat(events_dict[\"HHbbVV\"], \"VVFatJetParticleNetMass\", bb_masks[\"HHbbVV\"])\n",
+    "bbpnet = utils.get_feat(events_dict[\"HHbbVV\"], \"bbFatJetParticleNetMD_Xbb\", bb_masks[\"HHbbVV\"])\n",
+    "vvpart = utils.get_feat(events_dict[\"HHbbVV\"], \"VVFatJetParTMD_THWW4q\", bb_masks[\"HHbbVV\"])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "bbreg = (bbmsd < 50) * (bbpnetm > 50)\n",
+    "plt.title(\"Jets with Msd < 50 GeV and PNet Mass > 50 GeV\")\n",
+    "plt.hist(bbpnet[bbreg], np.linspace(0, 1, 50), histtype=\"step\", label=\"bb\")\n",
+    "plt.xlabel(\"Txbb\")\n",
+    "plt.show()"
+   ]
+  },
   {
    "attachments": {},
    "cell_type": "markdown",

diff --git a/src/HHbbVV/postprocessing/TrainBDT.py b/src/HHbbVV/postprocessing/TrainBDT.py
@@ -195,6 +195,19 @@ def main(args):
 
     data_dict = load_data(args.data_path, args.year, args.all_years)
 
+    for year, data in data_dict.items():
+        for key in training_keys:
+            print(
+                (
+                    f"{year} {key} Yield: "
+                    f'{np.sum(data[data["Dataset"] == key][weight_key])}, '
+                    "Number of Events: ",
+                    f'{len(data[data["Dataset"] == key])}, ',
+                )
+            )
+
+    return
+
     if not args.inference_only:
         training_data_dict = {
             year: data[