reformatting, check for CC mode

Author: jzuhone

kadi/commands/validate.py

@@ -722,7 +722,7 @@ def state_vals(self):
 class ValidateACISStatePower(ValidateSingleMsid):
     state_name = "dpa_power"
     msids = ["dpa_power"]
-    state_keys_extra = ["ccd_count", "clocking", "feps", "ccds", "fep_count", "simpos"]
+    state_keys_extra = ["ccd_count", "clocking", "feps", "ccds", "fep_count", "si_mode"]
     plot_attrs = PlotAttrs(title="DPA Power", ylabel="DPA Power (W)")
     min_violation_duration = 600.0  # seconds
     max_delta_val = 2.0  # W
@@ -744,24 +744,37 @@ def states(self):
 
     @functools.cached_property
     def state_vals(self):
-        istates = interpolate_states(self.states, self.times)
-        # create on-off states for FEPs
+        int_states = interpolate_states(self.states, self.times)
+
+        # create on-off states for FEPs and CCDs
         ccds = [f"I{i}" for i in range(4)] + [f"S{i}" for i in range(6)]
         ccdsfeps = defaultdict(list)
-        for row in istates:
+        for row in int_states:
             for i in range(6):
-                ccdsfeps[f"FEP{i}"].append(float(str(i) in row["feps"]))
-            for ccd in ccds:
-                ccdsfeps[ccd].append(float(str(ccd) in row["ccds"]))
+                ccdsfeps[f"FEP{i}"].append(str(i) in row["feps"])
+            for key in ccds:
+                ccdsfeps[key].append(key in row["ccds"])
         for key in ccdsfeps:
-            istates[key] = ccdsfeps[key]
-        df = istates.to_pandas()
-        keep_cols = list(ccdsfeps.keys())
-        XX = df.drop([col for col in istates.colnames if col not in keep_cols], axis=1)
+            int_states[key] = np.array(ccdsfeps[key], dtype="float64")
+
+        # check for continuous clocking mode
+        int_states["cc"] = np.char.startswith(int_states["si_mode"], "CC").astype(
+            "float64"
+        )
+        int_states["cc"] *= ~int_states["clocking"].astype("bool")
+
+        # Convert to pandas DataFrame, so we can use sklearn
+        df = int_states.to_pandas()
+        keep_cols = ["cc"] + list(ccdsfeps.keys())
+        XX = df.drop(
+            [col for col in int_states.colnames if col not in keep_cols], axis=1
+        )
         XX = self.scaler_X.fit_transform(XX.values)
         yy = self.model.predict(XX)
+
         # Inverse transform the predictions and actual values
         _state_vals = self.scaler_y.inverse_transform(yy.reshape(-1, 1)).flatten()
+
         return _state_vals
 
     def add_exclude_intervals(self):

kadi/scripts/create_dpa_power_model.py

@@ -3,6 +3,7 @@
 from collections import defaultdict
 
 import astropy.units as u
+import numpy as np
 from cheta import fetch_sci as fetch
 from cxotime import CxoTime
 from joblib import dump
@@ -59,11 +60,10 @@ def main():
     state_keys = [
         "ccd_count",
         "clocking",
-        "simpos",
         "feps",
         "ccds",
         "fep_count",
-        "power_cmd",
+        "si_mode",
     ]
 
     # Get commanded states
@@ -81,21 +81,25 @@ def main():
 
     logger.setLevel(args.log_level)
 
-    # create on-off states for FEPs
+    # create on-off states for FEPs and CCDs
     ccds = [f"I{i}" for i in range(4)] + [f"S{i}" for i in range(6)]
     ccdsfeps = defaultdict(list)
     for row in int_states:
         for i in range(6):
-            ccdsfeps[f"FEP{i}"].append(float(str(i) in row["feps"]))
-        for ccd in ccds:
-            ccdsfeps[ccd].append(float(str(ccd) in row["ccds"]))
+            ccdsfeps[f"FEP{i}"].append(str(i) in row["feps"])
+        for key in ccds:
+            ccdsfeps[key].append(key in row["ccds"])
     for key in ccdsfeps:
-        int_states[key] = ccdsfeps[key]
+        int_states[key] = np.array(ccdsfeps[key], dtype="float64")
+
+    # check for continuous clocking mode
+    int_states["cc"] = np.char.startswith(int_states["si_mode"], "CC").astype("float64")
+    int_states["cc"] *= ~int_states["clocking"].astype("bool")
 
     # Convert to pandas DataFrame, so we can use sklearn
     df = int_states.to_pandas()
 
-    model_keys = list(ccdsfeps.keys())
+    model_keys = ["cc"] + list(ccdsfeps.keys())
 
     # Separate into features and target
     X = df.drop([col for col in int_states.colnames if col not in model_keys], axis=1)