Merge pull request #354 from astrolabsoftware/master

merge master into branch for AL loop

.github/workflows/run_test.yml

@@ -42,8 +42,6 @@ jobs:
 
     - name: Run test suites
       run: |
-        pip uninstall -y supernnova
-        pip install git+https://github.com/supernnova/SuperNNova.git@6cfcb72009a734b26caa46eacb54d3e2dde9e2c3#egg=supernnova 
         pip uninstall -y actsnfink
         pip install git+https://github.com/emilleishida/fink_sn_activelearning.git@bf8d4e263e02d42781642f872f7bc030c24792bc#egg=actsnfink
 

fink_science/__init__.py

@@ -12,4 +12,4 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-__version__ = "4.8.1"
+__version__ = "5.1.0"

fink_science/agn/classifier.py

@@ -13,7 +13,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import pickle
+import joblib
 import fink_science.agn.kernel as k
 import fink_science.agn.feature_extraction as fe
 import os
@@ -27,7 +27,7 @@
 def load_classifier(source):
     """
     load the random forest classifier trained to recognize the AGN
-    on binary cases : AGNs vs non-AGNs  (pickle format).
+    on binary cases : AGNs vs non-AGNs  (joblib format).
 
     Parameters
     ----------
@@ -58,8 +58,7 @@ def load_classifier(source):
     elif source == 'ZTF':
         model_path = k.CLASSIFIER_ZTF
 
-    with open(model_path, "rb") as f:
-        clf = pickle.load(f)
+    clf = joblib.load(model_path)
 
     return clf
 

fink_science/agn/kernel.py

@@ -17,7 +17,7 @@
 from fink_science import __file__
 
 curdir = os.path.dirname(os.path.abspath(__file__))
-CLASSIFIER_ELASTICC = curdir + "/data/models/AGN_elasticc_alerts.pkl"
-CLASSIFIER_ZTF = curdir + "/data/models/AGN_binary.pkl"
+CLASSIFIER_ELASTICC = curdir + "/data/models/AGN_elasticc_alerts.joblib"
+CLASSIFIER_ZTF = curdir + "/data/models/AGN_binary.joblib"
 MINIMUM_POINTS = 4
 MAXFEV = 3000

fink_science/agn/unit_examples.py

@@ -1,11 +1,9 @@
 import pandas as pd
 import numpy as np
 import fink_science.agn.kernel as k
-import pickle
+import joblib
 
-
-with open(k.CLASSIFIER_ELASTICC, "rb") as f:
-    clf_unit = pickle.load(f)
+clf_unit = joblib.load(k.CLASSIFIER_ELASTICC)
 
 raw_ztf_unit = pd.DataFrame(
     {

fink_science/random_forest_snia/processor.py

@@ -1,4 +1,4 @@
-# Copyright 2019-2022 AstroLab Software
+# Copyright 2019-2023 AstroLab Software
 # Author: Julien Peloton
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -70,7 +70,12 @@ def apply_selection_cuts_ztf(
     return mask
 
 @pandas_udf(DoubleType(), PandasUDFType.SCALAR)
-def rfscore_sigmoid_full(jd, fid, magpsf, sigmapsf, cdsxmatch, ndethist, model=None) -> pd.Series:
+def rfscore_sigmoid_full(
+        jd, fid, magpsf, sigmapsf, cdsxmatch, ndethist,
+        min_rising_points=pd.Series([2]),
+        min_data_points=pd.Series([4]),
+        rising_criteria=pd.Series(['ewma']),
+        model=None) -> pd.Series:
     """ Return the probability of an alert to be a SNe Ia using a Random
     Forest Classifier (sigmoid fit).
 
@@ -88,6 +93,10 @@ def rfscore_sigmoid_full(jd, fid, magpsf, sigmapsf, cdsxmatch, ndethist, model=N
         Type of object found in Simbad (string)
     ndethist: Spark DataFrame Column
         Column containing the number of detection by ZTF at 3 sigma (int)
+    min_rising_points, min_data_points: int
+        Parameters from fink_sn_activelearning.git
+    rising_criteria: str
+        How to compute derivatives: ewma (default), or diff.
     model: Spark DataFrame Column, optional
         Path to the trained model. Default is None, in which case the default
         model `data/models/default-model.obj` is loaded.
@@ -139,16 +148,26 @@ def rfscore_sigmoid_full(jd, fid, magpsf, sigmapsf, cdsxmatch, ndethist, model=N
     +----------------+-----+
     <BLANKLINE>
 
-    # Note that we can also specify a model
+    # We can also specify fink_sn_activelearning parameters
     >>> args = [F.col(i) for i in what_prefix]
     >>> args += [F.col('cdsxmatch'), F.col('candidate.ndethist')]
-    >>> args += [F.lit(model_path_sigmoid)]
+    >>> args += [F.lit(2), F.lit(4), F.lit('ewma')]
     >>> df = df.withColumn('pIa', rfscore_sigmoid_full(*args))
 
     >>> df.filter(df['pIa'] > 0.5).count()
     6
 
-    >>> df.agg({"pIa": "max"}).collect()[0][0] < 1.0
+    # We can also specify a different model
+    >>> args = [F.col(i) for i in what_prefix]
+    >>> args += [F.col('cdsxmatch'), F.col('candidate.ndethist')]
+    >>> args += [F.lit(1), F.lit(3), F.lit('diff')]
+    >>> args += [F.lit(model_path_al_loop)]
+    >>> df = df.withColumn('pIaAL', rfscore_sigmoid_full(*args))
+
+    >>> df.filter(df['pIaAL'] > 0.5).count()
+    5
+
+    >>> df.agg({"pIaAL": "max"}).collect()[0][0] < 1.0
     True
     """
     mask = apply_selection_cuts_ztf(magpsf, ndethist, cdsxmatch)
@@ -171,7 +190,12 @@ def rfscore_sigmoid_full(jd, fid, magpsf, sigmapsf, cdsxmatch, ndethist, model=N
     flag = []
     for id in np.unique(pdf['SNID']):
         pdf_sub = pdf[pdf['SNID'] == id]
-        features = get_sigmoid_features_dev(pdf_sub)
+        features = get_sigmoid_features_dev(
+            pdf_sub,
+            min_rising_points=min_rising_points.values[0],
+            min_data_points=min_data_points.values[0],
+            rising_criteria=rising_criteria.values[0]
+        )
         if (features[0] == 0) or (features[6] == 0):
             flag.append(False)
         else:
@@ -183,14 +207,22 @@ def rfscore_sigmoid_full(jd, fid, magpsf, sigmapsf, cdsxmatch, ndethist, model=N
     # Make predictions
     probabilities = clf.predict_proba(test_features)
 
+    # pIa = 0.0 for objects that do not
+    # have both features non-zero.
+    probabilities[~flag] = [1.0, 0.0]
+
     # Take only probabilities to be Ia
     to_return = np.zeros(len(jd), dtype=float)
     to_return[mask] = probabilities.T[1]
 
     return pd.Series(to_return)
 
 @pandas_udf(StringType(), PandasUDFType.SCALAR)
-def extract_features_rf_snia(jd, fid, magpsf, sigmapsf, cdsxmatch, ndethist) -> pd.Series:
+def extract_features_rf_snia(
+        jd, fid, magpsf, sigmapsf, cdsxmatch, ndethist,
+        min_rising_points=pd.Series([2]),
+        min_data_points=pd.Series([4]),
+        rising_criteria=pd.Series(['ewma'])) -> pd.Series:
     """ Return the features used by the RF classifier.
 
     There are 12 features. Order is:
@@ -209,6 +241,10 @@ def extract_features_rf_snia(jd, fid, magpsf, sigmapsf, cdsxmatch, ndethist) ->
         Type of object found in Simbad (string)
     ndethist: Spark DataFrame Column
         Column containing the number of detection by ZTF at 3 sigma (int)
+    min_rising_points, min_data_points: int
+        Parameters from fink_sn_activelearning.git
+    rising_criteria: str
+        How to compute derivatives: ewma (default), or diff.
 
     Returns
     ----------
@@ -259,7 +295,12 @@ def extract_features_rf_snia(jd, fid, magpsf, sigmapsf, cdsxmatch, ndethist) ->
     test_features = []
     for id in np.unique(pdf['SNID']):
         pdf_sub = pdf[pdf['SNID'] == id]
-        features = get_sigmoid_features_dev(pdf_sub)
+        features = get_sigmoid_features_dev(
+            pdf_sub,
+            min_rising_points=min_rising_points.values[0],
+            min_data_points=min_data_points.values[0],
+            rising_criteria=rising_criteria.values[0]
+        )
         test_features.append(features)
 
     to_return_features = np.zeros((len(jd), len(RF_FEATURE_NAMES)), dtype=float)
@@ -414,5 +455,8 @@ def rfscore_sigmoid_elasticc(
     model_path_sigmoid = '{}/data/models/default-model_sigmoid.obj'.format(path)
     globs["model_path_sigmoid"] = model_path_sigmoid
 
+    model_path_al_loop = '{}/data/models/for_al_loop/model_20231009.pkl'.format(path)
+    globs["model_path_al_loop"] = model_path_al_loop
+
     # Run the test suite
     spark_unit_tests(globs)