Networkx example¶

This Example will show you step by step, how to use networkx graphs with a PHOTONAI Graph pipeline.

First: Necessary imports

In [1]:

            
                Copied!
                
from networkx.generators import barabasi_albert_graph, erdos_renyi_graph
from photonai.base import Hyperpipe, PipelineElement
from sklearn.model_selection import KFold
import networkx as nx
import numpy as np
from networkx.generators import barabasi_albert_graph, erdos_renyi_graph
from photonai.base import Hyperpipe, PipelineElement
from sklearn.model_selection import KFold
import networkx as nx
import numpy as np

Next: Generate graphs and labels

In [2]:

            
                Copied!
                
                    
                    
                
                

        
# create dataset of 1000 graphs
rr_graphs = [erdos_renyi_graph(20, .25) for _ in range(500)]
er_graphs = [erdos_renyi_graph(20, .2) for _ in range(500)]
graphs = rr_graphs + er_graphs

# we have to transform the networkx graphs into numpy graphs before passing them to photon
graphs = np.array([nx.to_numpy_array(g) for g in graphs])
graphs = np.expand_dims(graphs, axis=-1)

# create labels
rr_labels = [0] * 500
er_labels = [1] * 500
labels = rr_labels + er_labels
# create dataset of 1000 graphs
rr_graphs = [erdos_renyi_graph(20, .25) for _ in range(500)]
er_graphs = [erdos_renyi_graph(20, .2) for _ in range(500)]
graphs = rr_graphs + er_graphs

# we have to transform the networkx graphs into numpy graphs before passing them to photon
graphs = np.array([nx.to_numpy_array(g) for g in graphs])
graphs = np.expand_dims(graphs, axis=-1)

# create labels
rr_labels = [0] * 500
er_labels = [1] * 500
labels = rr_labels + er_labels

Setup Hyperpipe

In [3]:

            
                Copied!
                
                    
                    
                
                

        
# Design your Pipeline
my_pipe = Hyperpipe('networkx_example_pipe',
                    inner_cv=KFold(n_splits=2),
                    optimizer='sk_opt',
                    optimizer_params={'n_configurations': 25},
                    metrics=['accuracy', 'balanced_accuracy', 'recall', 'precision'],
                    best_config_metric='accuracy')

my_pipe.add(PipelineElement('GCNClassifier', feature_axis=0))
# Design your Pipeline
my_pipe = Hyperpipe('networkx_example_pipe',
                    inner_cv=KFold(n_splits=2),
                    optimizer='sk_opt',
                    optimizer_params={'n_configurations': 25},
                    metrics=['accuracy', 'balanced_accuracy', 'recall', 'precision'],
                    best_config_metric='accuracy')

my_pipe.add(PipelineElement('GCNClassifier', feature_axis=0))

Final step: Train your Hyperpipe

In [4]:

            
                Copied!
                
my_pipe.fit(graphs, labels)
my_pipe.fit(graphs, labels)

=====================================================================================================
PHOTONAI ANALYSIS: networkx_example_pipe
=====================================================================================================
Found 2 target classes: [0 1]

*****************************************************************************************************
Outer Cross validation Fold 1
*****************************************************************************************************
Did not find any hyperparameter to convert into skopt space.

/opt/hostedtoolcache/Python/3.9.16/x64/lib/python3.9/site-packages/photonai/optimization/scikit_optimize/sk_opt.py:169: UserWarning: Did not find any hyperparameter to convert into skopt space.
  warnings.warn(msg)
100%|██████████| 200/200 [00:22<00:00,  8.90it/s]
100%|██████████| 200/200 [00:21<00:00,  9.25it/s]

-----------------------------------------------------------------------------------------------------
BEST_CONFIG 
-----------------------------------------------------------------------------------------------------
{}
-----------------------------------------------------------------------------------------------------
VALIDATION PERFORMANCE
-----------------------------------------------------------------------------------------------------
+-------------------+-------------------+------------------+
|       METRIC      | PERFORMANCE TRAIN | PERFORMANCE TEST |
+-------------------+-------------------+------------------+
|      accuracy     |       0.7800      |      0.7737      |
| balanced_accuracy |       0.7802      |      0.7737      |
|       recall      |       0.7893      |      0.7815      |
|     precision     |       0.7779      |      0.7720      |
+-------------------+-------------------+------------------+

100%|██████████| 200/200 [00:41<00:00,  4.79it/s]

-----------------------------------------------------------------------------------------------------
TEST PERFORMANCE
-----------------------------------------------------------------------------------------------------
+-------------------+-------------------+------------------+
|       METRIC      | PERFORMANCE TRAIN | PERFORMANCE TEST |
+-------------------+-------------------+------------------+
|      accuracy     |       0.7775      |      0.7700      |
| balanced_accuracy |       0.7773      |      0.7704      |
|       recall      |       0.8015      |      0.7835      |
|     precision     |       0.7672      |      0.7525      |
+-------------------+-------------------+------------------+
*****************************************************************************************************
Finished all outer fold computations.

100%|██████████| 200/200 [00:52<00:00,  3.84it/s]

*****************************************************************************************************

ANALYSIS INFORMATION ================================================================================ 
Project Folder: /home/runner/work/photonai_graph/photonai_graph/documentation/docs/examples/networkx_example_pipe_results_2023-02-17_16-36-40,
Computation Time: 2023-02-17 16:36:40.365564 - 2023-02-17 16:38:09.836040
Duration: 0:01:29.470476
Optimized for: accuracy
Hyperparameter Optimizer: sk_opt

DUMMY RESULTS =======================================================================================
+-------------------+--+
| PERFORMANCE DUMMY |  |
+-------------------+--+
+-------------------+--+

AVERAGE PERFORMANCE ACROSS OUTER FOLDS ==============================================================
+-------------------+---------------+--------------+-----------+----------+
|    Metric Name    | Training Mean | Training Std | Test Mean | Test Std |
+-------------------+---------------+--------------+-----------+----------+
|      accuracy     |     0.7775    |     0.0      |    0.77   |   0.0    |
| balanced_accuracy |    0.777319   |     0.0      |  0.770393 |   0.0    |
|       recall      |    0.801489   |     0.0      |  0.783505 |   0.0    |
|     precision     |    0.767221   |     0.0      |  0.752475 |   0.0    |
+-------------------+---------------+--------------+-----------+----------+

BEST HYPERPARAMETER CONFIGURATION ===================================================================
{}


+--------+----------+-------------------+--------+-----------+----------------------------+
| fold # | accuracy | balanced_accuracy | recall | precision | Best Hyperparameter Config |
+--------+----------+-------------------+--------+-----------+----------------------------+
|   1*   |  0.7700  |       0.7704      | 0.7835 |   0.7525  |             {}             |
+--------+----------+-------------------+--------+-----------+----------------------------+

PHOTONAI 2.3.0 ======================================================================================
Your results are stored in /home/runner/work/photonai_graph/photonai_graph/documentation/docs/examples/networkx_example_pipe_results_2023-02-17_16-36-40
Go to https://explorer.photon-ai.com and upload your photonai_results.json for convenient result visualization! 
For more info and documentation visit https://www.photon-ai.com

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
File /opt/hostedtoolcache/Python/3.9.16/x64/lib/python3.9/site-packages/IPython/core/formatters.py:972, in MimeBundleFormatter.__call__(self, obj, include, exclude)
    969     method = get_real_method(obj, self.print_method)
    971     if method is not None:
--> 972         return method(include=include, exclude=exclude)
    973     return None
    974 else:

File /opt/hostedtoolcache/Python/3.9.16/x64/lib/python3.9/site-packages/sklearn/base.py:631, in BaseEstimator._repr_mimebundle_(self, **kwargs)
    629 output = {"text/plain": repr(self)}
    630 if get_config()["display"] == "diagram":
--> 631     output["text/html"] = estimator_html_repr(self)
    632 return output

File /opt/hostedtoolcache/Python/3.9.16/x64/lib/python3.9/site-packages/sklearn/utils/_estimator_html_repr.py:417, in estimator_html_repr(estimator)
    403 fallback_msg = (
    404     "In a Jupyter environment, please rerun this cell to show the HTML"
    405     " representation or trust the notebook. <br />On GitHub, the"
    406     " HTML representation is unable to render, please try loading this page"
    407     " with nbviewer.org."
    408 )
    409 out.write(
    410     f"<style>{style_with_id}</style>"
    411     f'<div id="{container_id}" class="sk-top-container">'
   (...)
    415     '<div class="sk-container" hidden>'
    416 )
--> 417 _write_estimator_html(
    418     out,
    419     estimator,
    420     estimator.__class__.__name__,
    421     estimator_str,
    422     first_call=True,
    423 )
    424 out.write("</div></div>")
    426 html_output = out.getvalue()

File /opt/hostedtoolcache/Python/3.9.16/x64/lib/python3.9/site-packages/sklearn/utils/_estimator_html_repr.py:151, in _write_estimator_html(out, estimator, estimator_label, estimator_label_details, first_call)
    149 """Write estimator to html in serial, parallel, or by itself (single)."""
    150 if first_call:
--> 151     est_block = _get_visual_block(estimator)
    152 else:
    153     with config_context(print_changed_only=True):

File /opt/hostedtoolcache/Python/3.9.16/x64/lib/python3.9/site-packages/sklearn/utils/_estimator_html_repr.py:127, in _get_visual_block(estimator)
    123 # check if estimator looks like a meta estimator wraps estimators
    124 if hasattr(estimator, "get_params"):
    125     estimators = [
    126         (key, est)
--> 127         for key, est in estimator.get_params(deep=False).items()
    128         if hasattr(est, "get_params") and hasattr(est, "fit")
    129     ]
    130     if estimators:
    131         return _VisualBlock(
    132             "parallel",
    133             [est for _, est in estimators],
    134             names=[f"{key}: {est.__class__.__name__}" for key, est in estimators],
    135             name_details=[str(est) for _, est in estimators],
    136         )

File /opt/hostedtoolcache/Python/3.9.16/x64/lib/python3.9/site-packages/sklearn/base.py:211, in BaseEstimator.get_params(self, deep)
    209 out = dict()
    210 for key in self._get_param_names():
--> 211     value = getattr(self, key)
    212     if deep and hasattr(value, "get_params") and not isinstance(value, type):
    213         deep_items = value.get_params().items()

AttributeError: 'Hyperpipe' object has no attribute 'best_config_metric'

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
File /opt/hostedtoolcache/Python/3.9.16/x64/lib/python3.9/site-packages/IPython/core/formatters.py:342, in BaseFormatter.__call__(self, obj)
    340     method = get_real_method(obj, self.print_method)
    341     if method is not None:
--> 342         return method()
    343     return None
    344 else:

File /opt/hostedtoolcache/Python/3.9.16/x64/lib/python3.9/site-packages/sklearn/base.py:625, in BaseEstimator._repr_html_inner(self)
    620 def _repr_html_inner(self):
    621     """This function is returned by the @property `_repr_html_` to make
    622     `hasattr(estimator, "_repr_html_") return `True` or `False` depending
    623     on `get_config()["display"]`.
    624     """
--> 625     return estimator_html_repr(self)

File /opt/hostedtoolcache/Python/3.9.16/x64/lib/python3.9/site-packages/sklearn/utils/_estimator_html_repr.py:417, in estimator_html_repr(estimator)
    403 fallback_msg = (
    404     "In a Jupyter environment, please rerun this cell to show the HTML"
    405     " representation or trust the notebook. <br />On GitHub, the"
    406     " HTML representation is unable to render, please try loading this page"
    407     " with nbviewer.org."
    408 )
    409 out.write(
    410     f"<style>{style_with_id}</style>"
    411     f'<div id="{container_id}" class="sk-top-container">'
   (...)
    415     '<div class="sk-container" hidden>'
    416 )
--> 417 _write_estimator_html(
    418     out,
    419     estimator,
    420     estimator.__class__.__name__,
    421     estimator_str,
    422     first_call=True,
    423 )
    424 out.write("</div></div>")
    426 html_output = out.getvalue()

File /opt/hostedtoolcache/Python/3.9.16/x64/lib/python3.9/site-packages/sklearn/utils/_estimator_html_repr.py:151, in _write_estimator_html(out, estimator, estimator_label, estimator_label_details, first_call)
    149 """Write estimator to html in serial, parallel, or by itself (single)."""
    150 if first_call:
--> 151     est_block = _get_visual_block(estimator)
    152 else:
    153     with config_context(print_changed_only=True):

File /opt/hostedtoolcache/Python/3.9.16/x64/lib/python3.9/site-packages/sklearn/utils/_estimator_html_repr.py:127, in _get_visual_block(estimator)
    123 # check if estimator looks like a meta estimator wraps estimators
    124 if hasattr(estimator, "get_params"):
    125     estimators = [
    126         (key, est)
--> 127         for key, est in estimator.get_params(deep=False).items()
    128         if hasattr(est, "get_params") and hasattr(est, "fit")
    129     ]
    130     if estimators:
    131         return _VisualBlock(
    132             "parallel",
    133             [est for _, est in estimators],
    134             names=[f"{key}: {est.__class__.__name__}" for key, est in estimators],
    135             name_details=[str(est) for _, est in estimators],
    136         )

File /opt/hostedtoolcache/Python/3.9.16/x64/lib/python3.9/site-packages/sklearn/base.py:211, in BaseEstimator.get_params(self, deep)
    209 out = dict()
    210 for key in self._get_param_names():
--> 211     value = getattr(self, key)
    212     if deep and hasattr(value, "get_params") and not isinstance(value, type):
    213         deep_items = value.get_params().items()

AttributeError: 'Hyperpipe' object has no attribute 'best_config_metric'

Out[4]:

Hyperpipe(name='networkx_example_pipe')