- Normal: Feature Selected Visualization
- PCA 2D Visualization
- PCA 3D Visualization
import matplotlib.pyplot as plt
import inanalysis_algo.algo_component as alc
from inanalysis_algo.utils import AlgoUtils
from sklearn.datasets import load_iris
import pandas as pd
import numpy as np
from sklearn.decomposition import PCA
from mpl_toolkits.mplot3d import Axes3D
def _pre_process(data, mode):
data = np.array(data)
if mode == 'PCA_3D':
pca = PCA(n_components=3)
data = pca.fit_transform(data)
elif mode == 'PCA_2D':
pca = PCA(n_components=2)
data = pca.fit_transform(data)
return data
def DBSCAN_model_preview(data, labels, x_axis_name='', y_axis_name='', mode=None):
# pre-processing
data = _pre_process(data, mode)
# no. of clusters
n_clusters = len(set(labels)) - (1 if -1 in labels else 0)
fig = plt.figure(1, figsize=(8, 6))
if mode == 'PCA_3D':
# 3D fig object
ax = Axes3D(fig, elev=-150, azim=110)
# scatter clusters
scatters = [ax.scatter(data[labels==i][:, 0], data[labels==i][:, 1], \
data[labels==i][:, 2], edgecolor='k', linewidth='1')\
for i in range(n_clusters)]
# scatter outliers
ax.scatter(data[labels==-1][:, 0], data[labels==-1][:, 1], data[labels==-1][:, 2], c='k')
else:
# scatter clusters
scatters = [plt.scatter(data[labels==i][:, 0], data[labels==i][:, 1], edgecolor='k', linewidth='1') \
for i in range(n_clusters)]
# scatter outliers
plt.scatter(data[labels==-1][:, 0], data[labels==-1][:, 1], c='k')
# legends and xy label
legends = ['Group '+str(i+1) for i in range(n_clusters)]+['Outlier']
plt.title('Iris model plot - mode = ' + str(mode))
plt.legend(legends)
if mode is None:
plt.xlabel(x_axis_name)
plt.ylabel(y_axis_name)
# save the fig.
plt.savefig(x_axis_name+' _ '+y_axis_name+'.png')
else:
plt.savefig(mode+'.png', dpi=80)
plt.show()
plt.close('all')
# eps, jpeg, jpg, pdf, pgf, png, ps, raw, rgba, svg, svgz, tif, tiff.def main():
do_something()
# normal plot
DBSCAN_model_preview(iris_data, labels, x_axis_name=x_axis_name, y_axis_name=y_axis_name)
# PCA - 2D plot
DBSCAN_model_preview(iris_data, labels, mode='PCA_2D')
# PCA - 3D plot
DBSCAN_model_preview(iris_data, labels, mode='PCA_3D')
if __name__ == '__main__':
main()DEBUG:inanalysis_algo.utils:Clustering DBSCAN Training
