[Python] How can you save/look at the information in a dendogram?

I am trying to analyze data to determine the outcome based on a dendrogram. The problem is that I have primarily 2 groups of data "H" and "U", both of which are analyzed together. At the end of the dendrogram, I need to know which area has more "H" and "U".

I tried using the dictionary information of the dendrogram, but when I opened the functions I noticed that at no point does it contain the information I put in through linkage.

The following is the code I'm using:

import numpy as np
import matplotlib.pyplot as plt
from scipy.cluster.hierarchy import dendrogram, linkage, set_link_color_palette
from scipy.cluster import hierarchy
import pandas as pd
from sklearn.decomposition import PCA

df=pd.read_csv(name+".csv", index_col=None, header=0)

normalized_df=(df-df.mean())/df.std()

pca=PCA(n_components=6, n_oversamples=6)
principalComponents = pca.fit_transform(df)
principalDF = pd.DataFrame(data = principalComponents, columns = ['principal component 1', 'principal component 2', 'principal component 3', 'principal component 4', 'principal component 5', 'principal component 6'])

#Adds the type (U/H) to the dataframe
finalDF = pd.concat([principalDF, full_df[['type']]], axis = 1)

#From here on is when I start to have problems
data = list(zip(finalDF['principal component 1'], finalDF['principal component 2']))

linkage_data = linkage(data, method='ward', metric='euclidean')
hierarchy.set_link_color_palette(['r','g','b','w'])
den=dendrogram(linkage_data)
plt.title("Attempt #1")
plt.show()


The dendrogram looks as expected. The problem is that I am unable to differentiate the type (H or U) of each point. As mentioned, I attempted to look at the properties of the dendrogram, such as icoord and dcoord, but I couldn't figure out what it meant.

The data itself I'm using is 6 different columns, with different numbers which is the data I normalize ((df-df.mean())/df.std), and then take the principal components.

Any advice?

Continue reading...