pynsn 1.0.7.dev6__tar.gz → 1.0.9__tar.gz

{pynsn-1.0.7.dev6 → pynsn-1.0.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: pynsn
-Version: 1.0.7.dev6
+Version: 1.0.9
 Summary: Creating Non-Symbolic Number Displays
 Author-email: Oliver Lindemann <lindemann@essb.eur.nl>
 Requires-Python: >=3.10
@@ -8,12 +8,12 @@ Description-Content-Type: text/markdown
 Classifier: License :: OSI Approved :: MIT License
 Requires-Dist: numpy>=2.1.3
 Requires-Dist: pandas>=2.2
+Requires-Dist: scipy>=1.14
+Requires-Dist: matplotlib>=3.9
 Requires-Dist: shapely>=2.0
 Requires-Dist: orjson>=3.10
 Requires-Dist: svgwrite>=1.4
 Requires-Dist: Pillow>=11.0
-Requires-Dist: scipy>=1.14 ; extra == "analysis"
-Requires-Dist: matplotlib>=3.9 ; extra == "analysis"
 Requires-Dist: Sphinx>=5.1 ; extra == "doc"
 Requires-Dist: sphinx-autodoc-typehints ; extra == "doc"
 Requires-Dist: sphinx-rtd-theme ; extra == "doc"
@@ -21,7 +21,6 @@ Requires-Dist: pygame>=2.5 ; extra == "pygame"
 Requires-Dist: pytest >=2.7.3 ; extra == "test"
 Project-URL: Documentation, https://github.com/lindemann09/PyNSN
 Project-URL: Source, https://github.com/lindemann09/PyNSN
-Provides-Extra: analysis
 Provides-Extra: doc
 Provides-Extra: pygame
 Provides-Extra: test

{pynsn-1.0.7.dev6 → pynsn-1.0.9}/pynsn/__init__.py

@@ -2,7 +2,7 @@
 # pylint: disable=C0413
 
 __author__ = "Oliver Lindemann <lindemann@cognitive-psychology.eu>"
-__version__ = '1.0.7-dev6'
+__version__ = '1.0.9'
 
 from sys import version_info as _python_version_info
 from ._misc import is_interactive_mode as _is_interactive_mode

{pynsn-1.0.7.dev6 → pynsn-1.0.9}/pynsn/_stimulus/properties.py

@@ -8,7 +8,7 @@ __author__ = "Oliver Lindemann <lindemann@cognitive-psychology.eu>"
 
 import enum
 from collections import OrderedDict
-from typing import Any, Dict, List, Tuple, Union
+from typing import Any, List, Tuple, Union
 
 import numpy as np
 import shapely

pynsn-1.0.9/pynsn/collections/__init__.py

@@ -0,0 +1,2 @@
+from ._coll_stim import CollectionStimuli, AbstractCollection
+from ._coll_stim_pairs import CollectionStimulusPairs

{pynsn-1.0.7.dev6 → pynsn-1.0.9}/pynsn/collections/_coll_stim.py

@@ -9,19 +9,17 @@ import pandas as pd
 
 from .. import _misc, defaults
 from .._stimulus import NSNStimulus
+from .._stimulus.properties import VPList, ensure_vp
 from ._abc_coll import AbstractCollection, ListNSNStimuli
 
-# FIXME not yet tested
-# TODO NSNStimuli have no names yet
-
 
 class CollectionStimuli(AbstractCollection):
     """Collection of NSNNumPairs"""
 
-    def __init__(self, lst: tp.Union[None, NSNStimulus] = None) -> None:
+    def __init__(self, lst: None | ListNSNStimuli = None) -> None:
 
         if isinstance(lst, tp.List):
-            for x in lst:
+            for x in lst:  # type check
                 if not isinstance(x, NSNStimulus):
                     raise RuntimeError(
                         f"lst must be a list of NSNStimulus and not {type(x)}")
@@ -35,7 +33,7 @@ class CollectionStimuli(AbstractCollection):
         """
 
         self.stimuli.append(stim)
-        self.reset_properties_dataframe()
+        self.reset_properties()
 
     def save(self, path: tp.Union[str, Path], zipped: bool = True):
         """Save the collection as json files organized in subfolder"""
@@ -76,8 +74,8 @@ class CollectionStimuli(AbstractCollection):
             rtn.append(NSNStimulus.from_dict(d))
         return rtn
 
-    def reset_properties_dataframe(self):
-        """reset dataframe of visual properties
+    def reset_properties(self):
+        """resets visual properties
 
         If the array `CollectionStimuli.stimuli` have been changed directly,
         the method needs to be called to ensure get valid property data
@@ -103,7 +101,19 @@ class CollectionStimuli(AbstractCollection):
 
         if len(self._prop_df) != len(self.stimuli):
             self._calc_properties()
-        # names = [x.name for x in self.stimuli]
-        # rtn = self._prop_df.copy()
-        # rtn["name"] = names
-        return self._prop_df
+
+        rtn = self._prop_df.copy()
+        rtn["names"] = [x.name for x in self.stimuli]
+        return rtn
+
+    def corr(self, properties: None | VPList = None) -> pd.DataFrame:
+        """Pairwise Pearson correlation between visual properties"""
+        if len(self._prop_df) != len(self.stimuli):
+            self._calc_properties()
+
+        df = self._prop_df
+        if properties is not None:
+            prop_names = [ensure_vp(p).name for p in properties]
+            df = df[prop_names]
+
+        return df.corr()

{pynsn-1.0.7.dev6 → pynsn-1.0.9}/pynsn/collections/_coll_stim_pairs.py

@@ -11,6 +11,7 @@ from .. import _misc, defaults
 from .._stimulus import NSNStimulus, NSNStimulusPair
 from .._stimulus.properties import ensure_vp, VPOrList
 from ._abc_coll import AbstractCollection, ListNSNStimPairs
+from ._coll_stim import CollectionStimuli
 
 
 class CollectionStimulusPairs(AbstractCollection):
@@ -19,7 +20,7 @@ class CollectionStimulusPairs(AbstractCollection):
     def __init__(self, lst: tp.Union[None, ListNSNStimPairs] = None) -> None:
 
         if isinstance(lst, tp.List):
-            for x in lst:
+            for x in lst:  # type check
                 if not isinstance(x, NSNStimulusPair):
                     raise RuntimeError(
                         f"lst must be a list of NSNStimulusPairs and not {type(x)}")
@@ -35,7 +36,7 @@ class CollectionStimulusPairs(AbstractCollection):
         """
 
         self.pairs.append(NSNStimulusPair(stim_a, stim_b, name))
-        self.reset_properties_dataframe()
+        self.reset_properties()
 
     def save(self, path: tp.Union[str, Path], zipped: bool = True):
         """Save the collection as json files organized in subfolder"""
@@ -81,7 +82,7 @@ class CollectionStimulusPairs(AbstractCollection):
                        name=c["name"])
         return rtn
 
-    def reset_properties_dataframe(self):
+    def reset_properties(self):
         """reset dataframe of visual properties
 
         If the array `CollectionStimulusPairs.pairs` have been changed directly,
@@ -158,3 +159,18 @@ class CollectionStimulusPairs(AbstractCollection):
             cols = ensure_vp(props).name
 
         return self._prop_df_a[cols] / self._prop_df_b[cols]
+
+    def collection_stimuli(self, stim_id: str) -> CollectionStimuli:
+        """
+        stim_id : {'a', 'b'}
+        """
+
+        if stim_id == "a":
+            rtn = [pair.stim_a for pair in self.pairs]
+        elif stim_id == "b":
+            rtn = [pair.stim_b for pair in self.pairs]
+        else:
+            raise ValueError(f"Unknown stimulus id '{stim_id}'. " +
+                             "Must be either 'a' or 'b'")
+
+        return CollectionStimuli(rtn)

{pynsn-1.0.7.dev6 → pynsn-1.0.9}/pynsn/collections/fit.py

@@ -15,6 +15,7 @@ def property_ratio_correlation(collection: CollectionStimulusPairs,
                                prop_a: _tp.Union[str, _VP],
                                prop_b: _tp.Union[None, str, _VP] = None,
                                max_corr: float = 0.01,
+                               adapt_stim: str = "both",
                                feedback: bool = True) -> _tp.Union[_tp.Tuple[float, float], float]:
 
     prop_a = _ensure_vp(prop_a)
@@ -30,13 +31,17 @@ def property_ratio_correlation(collection: CollectionStimulusPairs,
         if feedback:
             _sys.stdout.write(
                 f"fitting {i+1}/{n} {sp.name}                 \r")
-        _stim_fit.property_ratio(sp, prop_a, rnd_values[i, 0])
+        _stim_fit.property_ratio(
+            sp, prop_a, rnd_values[i, 0], adapt_stim=adapt_stim)
+
         if isinstance(prop_b, _VP):
-            _stim_fit.property_ratio(sp, prop_b, rnd_values[i, 1])
+            _stim_fit.property_ratio(
+                sp, prop_b, rnd_values[i, 1], adapt_stim=adapt_stim)
+
     if feedback:
         print(" "*70)
 
-    collection.reset_properties_dataframe()
+    collection.reset_properties()
     return target_correlations
 
 
@@ -68,7 +73,7 @@ def property_difference_correlation(collection: CollectionStimulusPairs,
     if feedback:
         print(" "*70)
 
-    collection.reset_properties_dataframe()
+    collection.reset_properties()
 
     return target_correlations
 

pynsn-1.0.9/pynsn/geometry.py

@@ -0,0 +1,51 @@
+
+from copy import deepcopy as _deepcopy
+
+import numpy as _np
+from numpy.typing import NDArray as _NDArray
+
+from ._stimulus.nsn_stimulus import NSNStimulus as _NSNStimulus
+
+
+def distances(shape_array: _NSNStimulus) -> _NDArray:
+    """Matrix with the distances between all shapes in the array"""
+    return _relation_matrix(shape_array, what=0)
+
+
+def dwithin(shape_array: _NSNStimulus, distance: float) -> _NDArray:
+    """Matrix with the booleans indicates dwithin between all shapes in the array"""
+    return _relation_matrix(shape_array, what=1, para=distance)
+
+
+def overlaps(shape_array: _NSNStimulus, min_distance: float | None = None) -> _NDArray:
+    if min_distance is None:
+        min_distance = shape_array.min_distance
+    else:
+        min_distance = 0
+
+    return dwithin(shape_array, distance=min_distance)
+
+
+def _relation_matrix(arr: _NSNStimulus, what: int, para: float = 0) -> _NDArray:
+    """helper function returning the relation between polygons
+    0 = distance
+    1 = dwithin
+    """
+    arr = _deepcopy(arr)
+    l = arr.n_shapes
+    rtn = _np.full((l, l), _np.nan)
+    for x in reversed(range(l)):
+        shape = arr.shape_pop(x)
+        if what == 0:
+            y = arr.distances(shape)
+        elif what == 1:
+            y = arr.dwithin(shape=shape, distance=para)
+        else:
+            raise RuntimeError("unknown function")
+
+        rtn[x, 0:x] = y
+
+    # make symmetric
+    i_lower = _np.triu_indices(l, 1)
+    rtn[i_lower] = rtn.T[i_lower]
+    return rtn

pynsn-1.0.9/pynsn/plots/__init__.py

@@ -0,0 +1,3 @@
+from ._collection_plots import (property_regression,
+                                property_difference_regression,
+                                property_ratio_regression)

pynsn-1.0.9/pynsn/plots/_collection_plots.py

@@ -0,0 +1,93 @@
+from math import ceil
+from typing import List, Tuple, Union
+
+import matplotlib.pyplot as plt
+import pandas as pd
+from scipy.stats import linregress
+
+from .._stimulus.properties import VP, ensure_vp, VPOrList
+from ..collections._coll_stim_pairs import CollectionStimulusPairs, CollectionStimuli
+
+
+def property_regression(stimuli: CollectionStimuli,
+                        dv: Union[str, VP],
+                        iv: VPOrList,
+                        figsize: Tuple[float, float] = (10, 8)):
+
+    if isinstance(iv, List):
+        iv_prop = [ensure_vp(p) for p in iv]
+    else:
+        iv_prop = [ensure_vp(iv)]
+    dv = ensure_vp(dv)
+
+    data = stimuli.property_dataframe()
+    cols = [p.name for p in iv_prop + [dv]]
+
+    return _regression_plots(data[cols], dv=dv,
+                             ivs=iv_prop, figsize=figsize)
+
+
+def property_difference_regression(stim_pairs: CollectionStimulusPairs,
+                                   dv: Union[str, VP],
+                                   iv: VPOrList,
+                                   figsize: Tuple[float, float] = (10, 8)):
+
+    if isinstance(iv, List):
+        iv_prop = [ensure_vp(p) for p in iv]
+    else:
+        iv_prop = [ensure_vp(iv)]
+    dv = ensure_vp(dv)
+
+    data = stim_pairs.property_differences(iv_prop + [dv])
+    return _regression_plots(data, dv=dv,
+                             ivs=iv_prop, figsize=figsize)
+
+
+def property_ratio_regression(stim_pairs: CollectionStimulusPairs,
+                              dv: Union[str, VP],
+                              iv: VPOrList,
+                              figsize: Tuple[float, float] = (10, 8)):
+
+    if isinstance(iv, List):
+        iv_prop = [ensure_vp(p) for p in iv]
+    else:
+        iv_prop = [ensure_vp(iv)]
+    dv = ensure_vp(dv)
+
+    data = stim_pairs.property_ratios(iv_prop + [dv])
+    return _regression_plots(data, dv=dv,
+                             ivs=iv_prop, figsize=figsize)
+
+
+def _regression_plots(data: pd.DataFrame,
+                      dv:  VP,
+                      ivs: List[VP],
+                      figsize: Tuple[float, float] = (10, 8)):
+
+    if dv in ivs:
+        raise ValueError(f"Dependent variable '{dv.name}' is also"
+                         " in list of independent variables.")
+    if len(ivs) == 1:
+        n_col = 1
+        n_row = 1
+    else:
+        n_col = 2
+        n_row = ceil(len(ivs)/2)
+
+    fig, axs = plt.subplots(n_row, n_col, figsize=figsize)
+    for i, p in enumerate(ivs):
+        _reg_plot(axs.flat[i], data, dv.name, p.name)
+
+    plt.tight_layout()
+    return fig
+
+
+def _reg_plot(ax: plt.Axes, df: pd.DataFrame, x: str, y: str):  # type: ignore
+    slope, intercept, r_value, _, _ = linregress(df[x], df[y])
+
+    ax.scatter(df[x], df[y])
+    # Add regression line
+    reg_line = slope * df[x] + intercept  # type: ignore
+    ax.plot(df[x], reg_line, color='green')
+    ax.set_title(f"{x}, {y}  (r={r_value:.2f})")
+    return

{pynsn-1.0.7.dev6 → pynsn-1.0.9}/pyproject.toml

@@ -12,13 +12,14 @@ dynamic = ["version", "description"]
 requires-python = ">=3.10"
 dependencies =  ["numpy>=2.1.3",
                     "pandas>=2.2",
+                    "scipy>=1.14",
+                    "matplotlib>=3.9",
                     "shapely>=2.0",
                     "orjson>=3.10",
                     "svgwrite>=1.4",
                     "Pillow>=11.0"]
 
 [project.optional-dependencies]
-analysis = ["scipy>=1.14", "matplotlib>=3.9"]
 pygame = ["pygame>=2.5"]
 #expyriment = ["expyriment>=0.10"]
 test = [

pynsn-1.0.7.dev6/pynsn/analysis/__init__.py

@@ -1 +0,0 @@
-from ._collection_stats import property_regression_sheet

pynsn-1.0.7.dev6/pynsn/analysis/_collection_stats.py

@@ -1,51 +0,0 @@
-from math import ceil
-from typing import List, Tuple, Union
-
-import matplotlib.pyplot as plt
-import pandas as pd
-from scipy.stats import linregress
-
-from .._stimulus.properties import VP, ensure_vp, VPOrList
-from ..collections._coll_stim_pairs import CollectionStimulusPairs
-
-
-def _regression_plot(ax: plt.Axes, df: pd.DataFrame, x: str, y: str):
-    slope, intercept, r_value, p_value, std_err = linregress(df[x], df[y])
-
-    ax.scatter(df[x], df[y])
-    # Add regression line
-    reg_line = slope * df[x] + intercept  # type: ignore
-    ax.plot(df[x], reg_line, color='green')
-    ax.set_title(f"{x}, {y}  (r={r_value:.2f})")
-    return
-
-
-def property_regression_sheet(stim_set: CollectionStimulusPairs,
-                              dv: Union[str, VP],
-                              iv: VPOrList,
-                              ratios: bool = False,
-                              figsize: Tuple[float, float] = (10, 8)):
-    if isinstance(iv, List):
-        iv = [ensure_vp(p) for p in iv]
-    else:
-        iv = [ensure_vp(iv)]
-    dv = ensure_vp(dv)
-
-    if ratios:
-        r = stim_set.property_ratios(iv + [dv])
-    else:
-        r = stim_set.property_differences(iv + [dv])
-
-    if len(iv) == 1:
-        n_col = 1
-        n_row = 1
-    else:
-        n_col = 2
-        n_row = ceil(len(iv)/2)
-
-    fig, axs = plt.subplots(n_row, n_col, figsize=figsize)
-    for i, p in enumerate(iv):
-        _regression_plot(axs.flat[i], r, dv.name, p.name)
-
-    plt.tight_layout()
-    return fig

pynsn-1.0.7.dev6/pynsn/collections/__init__.py

@@ -1,2 +0,0 @@
-from ._coll_stim import CollectionStimuli, AbstractCollection, ListNSNStimuli
-from ._coll_stim_pairs import CollectionStimulusPairs, ListNSNStimPairs

pynsn-1.0.7.dev6/pynsn/matrix/__init__.py

@@ -1,5 +0,0 @@
-"""calculate relation matrices """
-
-# pylint: disable=W0611
-
-from .matrix import distances, dwithin, overlaps

pynsn-1.0.7.dev6/pynsn/matrix/matrix.py

@@ -1,54 +0,0 @@
-
-from copy import deepcopy
-from typing import Optional
-
-import numpy as np
-from numpy.typing import NDArray
-
-from .._stimulus.nsn_stimulus import NSNStimulus
-from .._stimulus.shape_array import ShapeArray
-
-
-def distances(shape_array: ShapeArray) -> NDArray:
-    """Matrix with the distances between the shapes in the array"""
-    return _relation_matrix(shape_array, what=0)
-
-
-def dwithin(shape_array: ShapeArray, distance: float) -> NDArray:
-    """Matrix with the booleans indication dwithin between the shapes in the array"""
-    return _relation_matrix(shape_array, what=1, para=distance)
-
-
-def overlaps(shape_array: ShapeArray, min_distance: Optional[float] = None) -> NDArray:
-    if min_distance is None:
-        if isinstance(shape_array, NSNStimulus):
-            min_distance = shape_array.min_distance
-        else:
-            min_distance = 0
-
-    return dwithin(shape_array, distance=min_distance)
-
-
-def _relation_matrix(arr: ShapeArray, what: int, para: float = 0) -> NDArray:
-    """helper function returning the relation between polygons
-    0 = distance
-    1 = dwithin
-    """
-    arr = deepcopy(arr)
-    l = arr.n_shapes
-    rtn = np.full((l, l), np.nan)
-    for x in reversed(range(l)):
-        shape = arr.shape_pop(x)
-        if what == 0:
-            y = arr.distances(shape)
-        elif what == 1:
-            y = arr.dwithin(shape=shape, distance=para)
-        else:
-            raise RuntimeError("unknown function")
-
-        rtn[x, 0:x] = y
-
-    # make symmetric
-    i_lower = np.triu_indices(l, 1)
-    rtn[i_lower] = rtn.T[i_lower]
-    return rtn