BUILD: release Facet 1.2.0rc0

RELEASE_NOTES.rst

@@ -1,6 +1,23 @@
 Release Notes
 =============
 
+FACET 1.2
+---------
+
+FACET 1.2 introduces the ability to run simulations on a subsample of the data used to
+fit the underlying crossfit.
+One example where this can be useful is to use only a recent period of a time series as
+the baseline of a simulation.
+
+1.2.0
+~~~~~
+
+- API: new optional parameter `subsample` in method 
+  :meth:`.BaseUnivariateSimulator.simulate_feature` can be used to specify a subsample
+  to be used in the simulation (but simulating using a crossfit based on the full
+  sample)
+
+
 FACET 1.1
 ---------
 
@@ -29,6 +46,11 @@ by the :class:`.LearnerInspector`.
 FACET 1.0
 ---------
 
+1.0.3
+~~~~~
+
+- FIX: restrict package requirements to *gamma-pytools* 1.0.* and *sklearndf* 1.0.x, since FACET 1.0 is not compatible with *gamma-pytools* 1.1.* 
+
 1.0.2
 ~~~~~
 

azure-pipelines.yml

@@ -1,10 +1,10 @@
 trigger:
-  - 1.1.x
+  - 1.2.x
   - dev/*
   - release/*
 
 pr:
-  - 1.1.x
+  - 1.2.x
   - dev/*
   - release/*
 
@@ -17,20 +17,20 @@ schedules:
   displayName: Nightly full build
   branches:
     include:
-    - 1.1.x
+    - 1.2.x
 
 resources:
   repositories:
     - repository: sklearndf
       type: github
       endpoint: BCG-Gamma
       name: BCG-Gamma/sklearndf
-      ref: 1.1.x # todo - update to stable release
+      ref: 1.1.x
     - repository: pytools
       type: github
       endpoint: BCG-Gamma
       name: BCG-Gamma/pytools
-      ref: 1.1.x # todo - update to stable release
+      ref: 1.2.x
 
 variables:
   ${{ if not(startsWith(variables['Build.SourceBranch'], 'refs/pull/')) }}:

environment.yml

@@ -16,7 +16,7 @@ dependencies:
   - lightgbm >= 3.0
   - m2r = 0.2.*
   - matplotlib >= 3.3
-  - nbsphinx = 0.7.*
+  - nbsphinx ~= 0.8.5
   - numpy >= 1.16
   - pandas >= 1.0
   - pip >= 20

pyproject.toml

@@ -15,14 +15,14 @@ license = "Apache Software License v2.0"
 
 requires = [
     # direct requirements of gamma-facet
-    "gamma-pytools  >=1.1,<2",
+    "gamma-pytools  ~=1.1.2",
     "matplotlib     >=3.0,<3.4",
     "numpy          >=1.16,<1.21",
     "packaging      >=20",
     "pandas         >=0.24,<1.3",
     "scipy          >=1.2,<1.6",
     "shap           >=0.34,<0.40",
-    "sklearndf      >=1.1,<2",
+    "sklearndf      ~=1.1.0",
     # additional requirements of sklearndf
     "boruta         >=0.3",
     "lightgbm       >=3.0",
@@ -65,7 +65,7 @@ docs = [
     "sphinx == 3.4.*",
     "sphinx-autodoc-typehints == 1.11.*",
     "pydata-sphinx-theme == 0.4.*",
-    "nbsphinx == 0.7.*",
+    "nbsphinx ~= 0.8.5",
     "jupyter >= 1.0",
     "docutils",
     "xlrd == 1.2.*",
@@ -93,25 +93,25 @@ lightgbm       = "=3.0.*"
 scikit-learn   = "=0.21.*"
 # additional requirements of gamma-pytools
 joblib         = "=0.14.*"
-typing_inspect = "=0.4"
+typing_inspect = "=0.4.*"
 
 [build.matrix.max]
-gamma-pytools  = ">=1.1,<2"
-matplotlib     = "<3.4"
-numpy          = "<1.21"
+gamma-pytools  = "=1.1.*"
+matplotlib     = "=3.3.*"
+numpy          = "=1.20.*"
 packaging      = ">=20"
-pandas         = "<1.3"
+pandas         = "=1.2.*"
 python         = "=3.8.*"
-scipy          = "<1.6"
-shap           = ">=0.37,<0.40"
-sklearndf      = ">=1.1,<2"
+scipy          = "=1.5.*"
+shap           = "=0.39.*"
+sklearndf      = "=1.1.*"
 # additional requirements of sklearndf
 boruta         = ">=0.3"
 lightgbm       = ">=3.0"
 scikit-learn   = "=0.23.*"
 # additional requirements of gamma-pytools
-joblib         = "<1.1"
-typing_inspect = "<0.7"
+joblib         = "=1.0.*"
+typing_inspect = "=0.6.*"
 
 [tool.black]
 # quiet = "True"

src/facet/__init__.py

@@ -6,7 +6,7 @@
 """
 
 
-__version__ = "1.1.0"
+__version__ = "1.2.0rc0"
 
 __logo__ = (
     r"""

src/facet/data/partition/_partition.py

@@ -4,6 +4,7 @@
 
 import logging
 import math
+import operator as op
 from abc import ABCMeta, abstractmethod
 from typing import Any, Generic, Iterable, Optional, Sequence, Tuple, TypeVar
 
@@ -141,10 +142,10 @@ def __init__(
         if (
             lower_bound is not None
             and upper_bound is not None
-            and lower_bound >= upper_bound
+            and lower_bound > upper_bound
         ):
             raise ValueError(
-                f"arg lower_bound >= arg upper_bound: [{lower_bound}, {upper_bound})"
+                f"arg lower_bound > arg upper_bound: [{lower_bound}, {upper_bound})"
             )
 
         self._lower_bound = lower_bound
@@ -239,15 +240,17 @@ def fit(
         lower_bound = self._lower_bound
         upper_bound = self._upper_bound
 
-        if lower_bound is None:
-            lower_bound = np.nanquantile(values, q=0.025)
+        if lower_bound is None or upper_bound is None:
+            q3q1 = np.nanquantile(values, q=[0.75, 0.25])
+            inlier_range = op.sub(*q3q1) * 1.5  # iqr * 1.5
 
-        if upper_bound is None:
-            upper_bound = np.nanquantile(values, q=0.975)
-            if upper_bound < lower_bound:
-                upper_bound = lower_bound
-        elif upper_bound < lower_bound:
-            lower_bound = upper_bound
+            if lower_bound is None:
+                lower_bound = values[values >= q3q1[1] - inlier_range].min()
+
+            if upper_bound is None:
+                upper_bound = values[values <= q3q1[0] + inlier_range].max()
+
+        assert upper_bound >= lower_bound
 
         # calculate the step count based on the maximum number of partitions,
         # rounded to the next-largest rounded value ending in 1, 2, or 5