Merge pull request #52 from zStupan/main

Add visualization test

README.md

@@ -1,5 +1,5 @@
 <p align="center">
-  <img alt="logo" width="300" src="https://raw.githubusercontent.com/firefly-cpp/NiaARM/main/.github/logo/logo.png">
+  <img alt="logo" width="300" src="https://raw.githubusercontent.com/firefly-cpp/NiaARM/main/.github/images/logo.png">
 </p>
 
 ---
@@ -28,6 +28,7 @@ The current version includes (but is not limited to) the following functions:
 - searching for association rules,
 - providing output of mined association rules,
 - generating statistics about mined association rules.
+- visualization of association rules
 
 ## Installation
 
@@ -135,6 +136,25 @@ problem.rules.sort()
 problem.rules.to_csv('output.csv')
 ```
 
+### Visualization
+
+The framework currently supports the hill slopes visualization method presented in [4].
+
+```python
+from matplotlib import pyplot as plt
+from niaarm.visualize import hill_slopes
+
+# load data...
+# mine rules...
+
+hill_slopes(rule, dataset.transactions)
+plt.show()
+```
+<p>
+    <img alt="logo" src="https://raw.githubusercontent.com/firefly-cpp/NiaARM/main/.github/images/hill_slopes.png">
+</p>
+
+
 For a full list of examples see the [examples folder](https://github.com/firefly-cpp/NiaARM/tree/main/examples)
 in the GitHub repository.
 
@@ -190,6 +210,10 @@ Ideas are based on the following research papers:
 
 [3] I. Fister Jr., I. Fister [A brief overview of swarm intelligence-based algorithms for numerical association rule mining](https://arxiv.org/abs/2010.15524). arXiv preprint arXiv:2010.15524 (2020).
 
+[4] Fister, I. et al. (2020). Visualization of Numerical Association Rules by Hill Slopes.
+    In: Analide, C., Novais, P., Camacho, D., Yin, H. (eds) Intelligent Data Engineering and Automated Learning – IDEAL 2020.
+    IDEAL 2020. Lecture Notes in Computer Science(), vol 12489. Springer, Cham. https://doi.org/10.1007/978-3-030-62362-3_10
+
 ## License
 
 This package is distributed under the MIT License. This license can be found online at <http://www.opensource.org/licenses/MIT>.

niaarm/tests/test_data/README.md

@@ -3,3 +3,5 @@
 - The Abalone dataset is downloaded from https://archive.ics.uci.edu/ml/index.php.
 
 - The wiki_test_case dataset is composed from the first Table found in: https://en.wikipedia.org/wiki/Lift_(data_mining)
+
+- SpotyDataGen - Iztok Fister Jr., Grega Vrbančič, Lucija Brezočnik, Vili Podgorelec, Iztok Fister. [SportyDataGen: an online generator of endurance sports activity collections](http://iztok-jr-fister.eu/static/publications/225.pdf). In: CECIIS: Central European Conference on Information and Intelligent Systems, pp. 171-178, 2018.

niaarm/tests/test_visualization.py

@@ -0,0 +1,51 @@
+import os
+from unittest import TestCase
+import numpy as np
+from niaarm import Dataset, Feature, Rule
+from niaarm.visualize import hill_slopes, _ribbon
+
+
+class TestHillSlopes(TestCase):
+    def setUp(self):
+        sporty = Dataset(os.path.join(os.path.dirname(__file__), 'test_data', 'sportydatagen_generated.csv'))
+        self.transactions = sporty.transactions
+        antecedent = [
+            Feature('duration', 'float', 46.9530354402242, 65.87258373112326),
+            Feature('distance', 'float', 26.23676635110497, 53.29979966985809),
+            Feature('average_hr', 'float', 104.1241905565174, 141.39599912527686),
+            Feature('average_altitude', 'float', 17.587384648223903, 547.0467243284303),
+        ]
+
+        consequent = [
+            Feature('calories', 'float', 1096.8185894801436, 1209.0),
+            Feature('ascent', 'float', 0.0, 74.19297690681586),
+            Feature('descent', 'float', 0.0, 623.8817163897467),
+        ]
+
+        self.rule = Rule(antecedent, consequent)
+
+    def test_hill_slopes(self):
+        support = np.array([0.934286, 0.847143, 0.74, 0.561429, 0.244286, 0.225714, 0.00714286])
+        confidence = np.array([0.934286, 0.847095, 0.753823, 0.570336, 0.261468, 0.224771, 0.00458716])
+        length = np.array([1.32128, 1.19801, 1.05634, 0.800303, 0.357828, 0.318542, 0.00848896])
+        position = np.array([0.66064, 2.76738, 4.64837, 6.14703, 6.98756, 7.55052, 7.71862])
+
+        s = (length + support + confidence) / 2
+        a = np.sqrt(s * (s - length) * (s - support) * (s - confidence))
+        height = 2 * a / length
+        x = np.sqrt(support ** 2 - height ** 2)
+
+        vec = np.concatenate((-length / 2, -length / 2 + x, length / 2))
+        vec = (vec.reshape(3, 7) + position).T.reshape(len(vec))
+
+        height = np.concatenate((height, np.zeros(len(vec) - 7)))
+        height = np.reshape(height, (3, 7)).T.reshape(len(vec))
+        height = np.concatenate((np.zeros(1), height))[:len(vec)]
+
+        _, ax1 = hill_slopes(self.rule, self.transactions)
+        _, ax2 = _ribbon(vec, height)
+        ax1_xx, ax1_yy, ax1_zz, _ = ax1.collections[0]._vec
+        ax2_xx, ax2_yy, ax2_zz, _ = ax2.collections[0]._vec
+        self.assertTrue(np.allclose(ax1_xx, ax2_xx))
+        self.assertTrue(np.allclose(ax1_yy, ax2_yy))
+        self.assertTrue(np.allclose(ax1_zz, ax2_zz))

niaarm/visualize.py

@@ -79,7 +79,7 @@ def hill_slopes(rule, transactions):
         height = np.reshape(height, (3, num_features)).T.reshape(len(vec))
         height = np.concatenate((np.zeros(1), height))[:len(vec)]
 
-        fig, ax = ribbon(vec, height)
+        fig, ax = _ribbon(vec, height)
         ax.set_ylabel('Location')
         ax.set_yticks(range(num_features + 1))
         ax.set_yticklabels(range(num_features + 1))
@@ -88,7 +88,7 @@ def hill_slopes(rule, transactions):
         return fig, ax
 
 
-def ribbon(x, z, width=0.5):
+def _ribbon(x, z, width=0.5):
     fig, ax = plt.subplots(subplot_kw={'projection': '3d'})
 
     xi = np.linspace(x[:-1], x[1:], num=100, axis=1).flatten()