Refactor and test summarize design

src/fmu/tools/sensitivities/_designsummary.py

@@ -3,6 +3,12 @@
 import pandas as pd
 
 
+def _get_sensitivity_type(senscase: str) -> str:
+    """Determine sensitivity type based on the case name"""
+    sensitivity_types = {"p10_p90": "mc", "ref": "ref", "skip": "skip"}
+    return sensitivity_types.get(senscase.lower(), "scalar")
+
+
 def summarize_design(filename, sheetname="DesignSheet01"):
     """
      Summarizes the design set up for one by one sensitivities
@@ -34,7 +40,22 @@ def summarize_design(filename, sheetname="DesignSheet01"):
 
     """
 
-    # Initialisation of dataframe to store results
+    # Read design matrix
+    if str(filename).endswith(".xlsx"):
+        dgn = pd.read_excel(filename, sheetname, engine="openpyxl")
+        # Drop empty rows or columns that have been read in
+        # due to having background colour/formatting
+        dgn.dropna(axis=0, how="all", inplace=True)
+        dgn = dgn.loc[:, ~dgn.columns.str.contains("^Unnamed")]
+    elif str(filename).endswith(".csv"):
+        dgn = pd.read_csv(filename)
+    else:
+        raise ValueError(
+            "Design matrix must be on Excel or csv format"
+            " and filename must end with .xlsx or .csv"
+        )
+
+    # Initialize results DataFrame with same columns
     designsummary = pd.DataFrame(
         columns=[
             "sensno",
@@ -48,126 +69,59 @@ def summarize_design(filename, sheetname="DesignSheet01"):
             "endreal2",
         ]
     )
-    sensno = 0
-    startreal1 = 0
-    endreal1 = 0
-
-    # Read design matrix and find realisation numbers for each sensitivity
-    if str(filename).endswith(".xlsx"):
-        dgn = pd.read_excel(filename, sheetname, engine="openpyxl")
-
-        # Drop empty rows or columns that have been read in
-        # due to having background colour/formatting
-
-        dgn.dropna(axis=0, how="all", inplace=True)
-        dgn = dgn.loc[:, ~dgn.columns.str.contains("^Unnamed")]
 
-    elif str(filename).endswith(".csv"):
-        dgn = pd.read_csv(filename)
+    # Get unique sensitivity names in order of appearance
+    sensnames = dgn["SENSNAME"].unique()
 
-    else:
-        raise ValueError(
-            "Design matrix must be on Excel or csv format"
-            " and filename must end with .xlsx or .csv"
-        )
-    sensname = dgn.loc[0]["SENSNAME"]
-    casename1 = dgn.loc[0]["SENSCASE"]
-    if casename1.lower() == "p10_p90":
-        senstype = "mc"
-    elif casename1.lower() == "ref":
-        senstype = "ref"
-    else:
-        senstype = "scalar"
-
-    currentsensname = sensname
-    currentsenscase = casename1
-    # starting with first case
-    secondcase = False
-    casename2 = None
-    startreal2 = None
-    endreal2 = None
-
-    for row in dgn.itertuples():
-        if currentsensname == row.SENSNAME and currentsenscase == row.SENSCASE:
-            if secondcase is True:
-                endreal2 = row.REAL
-            else:
-                endreal1 = row.REAL
-        elif currentsensname == row.SENSNAME:
-            secondcase = True
-            startreal2 = row.REAL
-            endreal2 = row.REAL
-            casename2 = row.SENSCASE
-            currentsensname = row.SENSNAME
-            currentsenscase = casename2
-        else:
-            if senstype != "skip":
-                if secondcase is True:
-                    designsummary.loc[sensno] = [
-                        sensno,
-                        sensname,
-                        senstype,
-                        casename1,
-                        startreal1,
-                        endreal1,
-                        casename2,
-                        startreal2,
-                        endreal2,
-                    ]
-                    sensno += 1
-                else:
-                    designsummary.loc[sensno] = [
-                        sensno,
-                        sensname,
-                        senstype,
-                        casename1,
-                        startreal1,
-                        endreal1,
-                        None,
-                        None,
-                        None,
-                    ]
-                    sensno += 1
-            secondcase = False
-            startreal1 = row.REAL
-            endreal1 = row.REAL
-
-            casename1 = row.SENSCASE
-            sensname = row.SENSNAME
-            currentsenscase = casename1
-            currentsensname = sensname
-            if row.SENSCASE.lower() == "p10_p90":
-                senstype = "mc"
-            elif row.SENSCASE.lower() == "skip":
-                senstype = "skip"
-            else:
-                senstype = "scalar"
-
-    # For last row
-    if senstype != "skip":
-        if secondcase is True:
-            designsummary.loc[sensno] = [
-                sensno,
-                sensname,
-                senstype,
-                casename1,
-                startreal1,
-                endreal1,
-                casename2,
-                startreal2,
-                endreal2,
-            ]
+    sensno = 0
+    for sensname in sensnames:
+        sens_group = dgn[dgn["SENSNAME"] == sensname].copy()
+        # Get cases in order of appearance
+        cases = sens_group.drop_duplicates("SENSCASE")[["SENSCASE"]].values.flatten()
+
+        # Skip if first case type is 'skip'
+        # The "skip" option was added when creating tornado plots
+        # with calc_tornadoinput.
+        # It allowed manual exclusion of sensitivities from tornado plots by changing
+        # SENSCASE to "skip" in the design matrix after running the experiment.
+        # This would exclude the sensitivity from both the
+        # *designsummary table and the tornado plot.
+        # calc_tornadoinput is deprecated so this can be removed once
+        # calc_tornadoinput is removed.
+        senstype = _get_sensitivity_type(cases[0])
+        if senstype == "skip":
+            continue
+
+        # First case
+        case1_data = sens_group[sens_group["SENSCASE"] == cases[0]]
+        casename1 = cases[0]
+        startreal1 = case1_data["REAL"].min()
+        endreal1 = case1_data["REAL"].max()
+
+        # Handle second case if it exists
+        if len(cases) > 1:
+            case2_data = sens_group[sens_group["SENSCASE"] == cases[1]]
+            casename2 = cases[1]
+            startreal2 = case2_data["REAL"].min()
+            endreal2 = case2_data["REAL"].max()
         else:
-            designsummary.loc[sensno] = [
-                sensno,
-                sensname,
-                senstype,
-                casename1,
-                startreal1,
-                endreal1,
-                None,
-                None,
-                None,
-            ]
+            casename2 = None
+            startreal2 = None
+            endreal2 = None
+
+        # Add row to results
+        designsummary.loc[sensno] = [
+            sensno,
+            sensname,
+            senstype,
+            casename1,
+            startreal1,
+            endreal1,
+            casename2,
+            startreal2,
+            endreal2,
+        ]
+
+        sensno += 1
 
     return designsummary

tests/sensitivities/test_calc_tornado.py

@@ -15,20 +15,44 @@ def test_designsummary():
     )
     # checking dimensions and some values in summary of design matrix
     assert snorrebergdesign.shape == (7, 9)
+
+    assert (
+        snorrebergdesign.columns
+        == [
+            "sensno",
+            "sensname",
+            "senstype",
+            "casename1",
+            "startreal1",
+            "endreal1",
+            "casename2",
+            "startreal2",
+            "endreal2",
+        ]
+    ).all()
     assert snorrebergdesign["sensname"][0] == "rms_seed"
+    assert snorrebergdesign["senstype"][0] == "mc"
+    assert snorrebergdesign["casename1"][0] == "P10_P90"
+    assert snorrebergdesign["startreal1"][0] == 0
+    assert snorrebergdesign["endreal1"][0] == 9
+    assert snorrebergdesign["casename2"][0] is None
+    assert snorrebergdesign["startreal2"][0] is None
+    assert snorrebergdesign["endreal2"][0] is None
+
+    assert snorrebergdesign["sensname"][6] == "relp_go"
+    assert snorrebergdesign["senstype"][6] == "scalar"
+    assert snorrebergdesign["casename1"][6] == "lc"
+    assert snorrebergdesign["startreal1"][6] == 90
+    assert snorrebergdesign["endreal1"][6] == 99
+    assert snorrebergdesign["casename2"][6] == "hc"
     assert snorrebergdesign["startreal2"][6] == 100
     assert snorrebergdesign["endreal2"][6] == 109
+
     assert snorrebergdesign["endreal1"].sum() == 333
 
     # Test same also when design matrix is in .csv format
     designcsv = summarize_design(TESTDATA / "distributions/design.csv")
-
-    # checking dimensions and some values in summary of design matrix
-    assert designcsv.shape == (7, 9)
-    assert designcsv["sensname"][0] == "rms_seed"
-    assert designcsv["startreal2"][6] == 100
-    assert designcsv["endreal2"][6] == 109
-    assert designcsv["endreal1"].sum() == 333
+    assert snorrebergdesign.equals(designcsv)
 
 
 def test_calc_tornadoinput():