The "first" version.

README.md

@@ -20,10 +20,75 @@ pip install soinn
 python -m build
 `
 
+
+
+## API
+
+`learn(data: object = None, dead_age: int = 100, lambda: int = 100, noise: float = 0.5, num_layer: int = 1) -> object`
+
+**input** : numpy array of shape (n, dim)
+
+**dead_age ** : the dead age of edges during the learning process, default 100
+
+**lamda** : the number of iterations before removing noise, default 100
+
+**noise** : the ratio of noise in input data, default 0.5
+
+**num_layer** : the number of soinn layer, default 1
+
+**return** : return a one dimensional numpy array contaning all learned cendrioids
+
+
+
 ## Run demo
-After install soinn, you can run a demo to show how soinn works.
+Here is a simple code using SOINN to learn cendroids of the input data
+
+
+
+`from scipy.io import loadmat
+
+import matplotlib.pyplot as plt
+
+import soinn
+
+
+
+data = loadmat("train.mat")['train']
+
+print('Load data with shape', data.shape)
+
+
+
+ax=plt.subplot(121)
+
+ax.set_title('origin data')
+
+plt.plot(data[:,0], data[:,1], '.')
+
+
+
+clus = soinn.learn(data)
+
+print('soinn learned clusters with shape', clus.shape)
+
+ax=plt.subplot(122)
+
+ax.set_title('learned cendroids')
+
+plt.plot(clus[:,0], clus[:,1], '.')
+
+
+
+plt.show()
+
+`
+
+You can run it by:
+
 `
 cd demo && python demo.py
 `
+
 You can get the following result
 ![image](https://github.com/uestc-lfs/soinn/blob/master/demo/result.png)
+

setup.py

@@ -9,7 +9,7 @@
 from setuptools import Extension, setup
 from setuptools.command.build_ext import build_ext
 
-__version__ = '0.2.0'
+__version__ = '0.2.1'
 
 
 include_dirs = [