TDA and Statistics using Gudhi Python Library

Part 4 - Classification using Persistence Homology

In [1]:
from IPython.display import Image
Image("SlidesGudhi/GeneralPipeLine_ML.png",width= 1000)
Out[1]:
In [ ]:
import pandas as pd
import numpy as np
import pickle as pickle
import gudhi as gd
from persistence_graphical_tools_Bertrand import *
%matplotlib inline

Load the data

In [ ]:
f = open("data_acc","rb")
data = pickle.load(f)
f.close()
data_A = data[0]
data_B = data[1] 
data_C = data[2]
label = data[3]
print(label)
data_A_sample = data_A[0]

Persistence Landscapes

The persistence landscape has been introduced Bubenik etal. JMLR 2015 as an alternative to persistence diagrams. This approach aims at representing topological features in an Hilbert space, for which statistical learning methods can be directly applied.

Note that many other alternatives have been proposed: silhouettes, persistence images, cumulative peristence intensity function etc. Coming soon in Gudhi...

In [3]:
Image("SlidesGudhi/Landscapes.png",width = 800)
Out[3]: