In [4]:
import pandas as pd
import numpy as np
import os
from sklearn.preprocessing import MinMaxScaler,StandardScaler,normalize
import matplotlib.pyplot as plt 
#importing the header files 
from collections import Counter

from sklearn.manifold import TSNE
import time
import sklearn
from sklearn.decomposition import PCA 

from sklearn import mixture


from sklearn.preprocessing import MinMaxScaler,StandardScaler,LabelEncoder
from sklearn.model_selection import GridSearchCV

import plotly.express as px
from collections import Counter 
from plotly.subplots import make_subplots
import plotly.graph_objects as go
import umap
from audio_results_util import load_dataset,plotClusters,encodedLabels,plotData3d,plotData2d,plotAllClusterModels,actualDistribution
%matplotlib inline
In [5]:
train_data,true_labels,true_encoded_labels=load_dataset('Audio_features_vowel/logMelSpec_muslim_features_vowel.csv')
In [6]:
train_data.shape
Out[6]:
(240, 128)
In [7]:
results_df=pd.read_csv('audio-results4/logMelSpec_muslim_features_vowel.csv')
results_df
Out[7]:
Unnamed: 0 n_clusters ARI AMI H C V FM A R P silhouette calinski davies
0 K-Means 5 0.002200 2.547598e-02 0.067865 0.044364 0.053654 0.322804 0.212500 0.285014 0.223919 0.099611 41.453568 2.207506
1 Agglomerative clustering 3 -0.009430 2.239266e-03 0.017141 0.038219 0.023668 0.595885 0.566667 0.185196 0.198374 0.259431 39.873813 1.148859
2 Birch 158 0.002524 3.430180e-02 0.823178 0.169803 0.281533 0.057021 0.004167 0.000040 0.003165 0.032680 5.587613 0.782031
3 DBSCAN 2 0.029059 3.489222e-02 0.038867 0.106604 0.056965 0.636879 0.600000 0.150943 0.111628 0.176915 10.795369 2.746961
4 Mean-shift 2 0.010322 1.530699e-02 0.015666 0.158000 0.028505 0.683867 0.650000 0.196226 0.132766 0.174987 5.762039 1.657554
5 Optics 1 0.000000 -5.797594e-17 0.000000 1.000000 0.000000 0.694318 0.662500 0.200000 0.132500 -1.000000 -1.000000 -1.000000
6 Gaussian-mixture 5 0.026097 1.307371e-02 0.051033 0.037006 0.042902 0.382041 0.137500 0.097363 0.147301 0.104725 39.759095 2.159768
In [8]:
actualDistribution(train_data,true_labels,true_encoded_labels)
plotAllClusterModels(train_data,'audio-results4/logMelSpec_muslim_features_vowel.csv')

Actual Distribution of the labels in the dataset: --> 
Counter({'a': 159, 'u': 47, 'cantdecide': 13, 'misaligned': 13, 'other': 8})
Counter({0: 159, 4: 47, 1: 13, 2: 13, 3: 8})

audio-results4/logMelSpec_muslim_features_vowel.csv_kmeans_labels.npy
K-Means
predicted_labels--> Counter({1: 66, 4: 55, 2: 52, 0: 43, 3: 24})
 2D representation

 3D representation

Agglomerative
predicted_labels--> Counter({0: 205, 1: 32, 2: 3})
 2D representation

 3D representation

Birch
Too many labels to show
DBSCAN
predicted_labels--> Counter({0: 215, -1: 22, 1: 3})
 2D representation

 3D representation

OPTICS
predicted_labels--> Counter({0: 240})
 2D representation

 3D representation

MEAN-SHIFT
predicted_labels--> Counter({0: 235, 1: 5})
 2D representation

 3D representation

Gaussian-Mixture
predicted_labels--> Counter({2: 100, 3: 51, 0: 50, 4: 29, 1: 10})
 2D representation

 3D representation
In [ ]:

PCA transformed data

In [ ]:
In [9]:
results_df=pd.read_csv('audio-results4/logMelSpec_muslim_features_vowel.csv-pca.csv')
results_df
Out[9]:
Unnamed: 0 n_clusters ARI AMI H C V FM A R P silhouette calinski davies
0 K-Means 5 0.004415 0.018267 0.059036 0.038486 0.046596 0.323117 0.175000 0.243956 0.177445 0.260406 132.380528 1.119692
1 Agglomerative clustering 2 0.005602 -0.007056 0.004647 0.025510 0.007862 0.665301 0.633333 0.191195 0.132751 0.507387 69.518393 0.684968
2 Birch 46 0.012161 0.046571 0.386636 0.113330 0.175282 0.154923 0.033333 0.001094 0.017391 0.282222 85.887384 0.738876
3 DBSCAN 2 -0.019487 0.011426 0.026591 0.033791 0.029762 0.466281 0.066667 0.122722 0.180180 -0.035591 9.451787 2.743165
4 Mean-shift 3 0.000566 0.017329 0.029958 0.060631 0.040102 0.593788 0.570833 0.214707 0.217734 0.342121 90.849797 1.047782
5 Optics 2 0.002223 -0.007566 0.012680 0.030276 0.017874 0.619345 0.041667 0.022254 0.146780 -0.294473 3.118736 2.328787
6 Gaussian-mixture 5 -0.024460 0.020666 0.059427 0.045061 0.051257 0.373637 0.370833 0.204426 0.216209 0.218962 117.703745 1.145961
In [10]:
pca_transformed=PCA(n_components=3).fit_transform(train_data)
pca_transformed=pd.DataFrame(pca_transformed)
actualDistribution(pca_transformed,true_labels,true_encoded_labels)
plotAllClusterModels(pca_transformed,'audio-results4/logMelSpec_muslim_features_vowel.csv-pca')

Actual Distribution of the labels in the dataset: --> 
Counter({'a': 159, 'u': 47, 'cantdecide': 13, 'misaligned': 13, 'other': 8})
Counter({0: 159, 4: 47, 1: 13, 2: 13, 3: 8})

audio-results4/logMelSpec_muslim_features_vowel.csv-pca_kmeans_labels.npy
K-Means
predicted_labels--> Counter({1: 65, 2: 51, 0: 51, 3: 49, 4: 24})
 2D representation

 3D representation

Agglomerative
predicted_labels--> Counter({0: 229, 1: 11})
 2D representation

 3D representation

Birch
Too many labels to show
DBSCAN
predicted_labels--> Counter({-1: 122, 1: 111, 0: 7})
 2D representation

 3D representation

OPTICS
predicted_labels--> Counter({-1: 213, 1: 16, 0: 11})
 2D representation

 3D representation

MEAN-SHIFT
predicted_labels--> Counter({0: 203, 1: 30, 2: 7})
 2D representation

 3D representation

Gaussian-Mixture
predicted_labels--> Counter({0: 118, 4: 43, 1: 36, 3: 33, 2: 10})
 2D representation

 3D representation
In [ ]:

t-SNE transformed data

In [11]:
results_df=pd.read_csv('audio-results4/logMelSpec_muslim_features_vowel.csv-tsne.csv')
results_df
Out[11]:
Unnamed: 0 n_clusters ARI AMI H C V FM A R P silhouette calinski davies
0 K-Means 5 0.002649 8.649036e-03 0.047400 0.030124 0.036837 0.310762 0.212500 0.203251 0.198384 0.242287 85.409443 1.125963
1 Agglomerative clustering 240 0.000000 -6.363593e-14 1.000000 0.186350 0.314157 0.000000 0.004167 0.000026 0.004167 -1.000000 -1.000000 -1.000000
2 Birch 240 0.000000 -6.363593e-14 1.000000 0.186350 0.314157 0.000000 0.008333 0.000115 0.008333 -1.000000 -1.000000 -1.000000
3 DBSCAN 2 -0.023686 6.663225e-03 0.018043 0.045703 0.025872 0.600577 0.566667 0.142558 0.108453 -0.062689 0.949784 7.920837
4 Mean-shift 2 -0.005764 -4.945772e-03 0.001684 0.063718 0.003281 0.689262 0.658333 0.198742 0.132218 0.534289 8.247693 0.332222
5 Optics 1 0.000000 -5.797594e-17 0.000000 1.000000 0.000000 0.694318 0.662500 0.200000 0.132500 -1.000000 -1.000000 -1.000000
6 Gaussian-mixture 5 0.009037 1.668214e-02 0.057391 0.036678 0.044754 0.318527 0.183333 0.235313 0.207199 0.223369 78.861885 1.185786
In [12]:
tsne_transformed=TSNE(n_components=3, n_jobs=-1).fit_transform(train_data)
tsne_transformed=pd.DataFrame(tsne_transformed)
actualDistribution(tsne_transformed,true_labels,true_encoded_labels)

plotAllClusterModels(train_data,'audio-results4/logMelSpec_muslim_features_vowel.csv-tsne')

Actual Distribution of the labels in the dataset: --> 
Counter({'a': 159, 'u': 47, 'cantdecide': 13, 'misaligned': 13, 'other': 8})
Counter({0: 159, 4: 47, 1: 13, 2: 13, 3: 8})

audio-results4/logMelSpec_muslim_features_vowel.csv-tsne_kmeans_labels.npy
K-Means
predicted_labels--> Counter({0: 54, 3: 48, 1: 48, 2: 46, 4: 44})
 2D representation

 3D representation

Agglomerative
Too many labels to show
Birch
Too many labels to show
DBSCAN
predicted_labels--> Counter({0: 209, -1: 30, 1: 1})
 2D representation

 3D representation

OPTICS
predicted_labels--> Counter({0: 240})
 2D representation

 3D representation

MEAN-SHIFT
predicted_labels--> Counter({0: 239, 1: 1})
 2D representation

 3D representation

Gaussian-Mixture
predicted_labels--> Counter({3: 57, 2: 55, 1: 48, 0: 41, 4: 39})
 2D representation

 3D representation
In [ ]:

Umap transformed data

In [13]:
results_df=pd.read_csv('audio-results4/logMelSpec_muslim_features_vowel.csv-umap.csv')
results_df
Out[13]:
Unnamed: 0 n_clusters ARI AMI H C V FM A R P silhouette calinski davies
0 K-Means 5 0.018758 0.018493 0.059656 0.038091 0.046495 0.325725 0.154167 0.155904 0.157838 0.413348 275.045396 0.882485
1 Agglomerative clustering 2 0.035808 0.010670 0.017217 0.027502 0.021177 0.532795 0.470833 0.170392 0.144561 0.422720 225.045855 0.891764
2 Birch 17 0.008194 0.027060 0.188880 0.069359 0.101461 0.181885 0.091667 0.042251 0.086190 0.408973 264.023639 0.819391
3 DBSCAN 13 -0.052288 0.016954 0.130299 0.068498 0.089792 0.320376 0.025000 0.002695 0.053571 0.009459 39.929345 0.952941
4 Mean-shift 3 -0.000997 -0.007293 0.010235 0.009733 0.009978 0.407904 0.312500 0.235607 0.162040 0.358024 220.848124 1.066792
5 Optics 2 -0.064487 0.007065 0.024402 0.036197 0.029151 0.519048 0.087500 0.022013 0.140000 0.137265 78.253097 1.077156
6 Gaussian-mixture 5 0.023662 0.025133 0.067936 0.043367 0.052940 0.329456 0.270833 0.221099 0.245030 0.413091 273.197966 0.896744
In [14]:
umap_transformed= umap.UMAP(random_state=42,n_components=3).fit_transform(train_data)
umap_transformed=pd.DataFrame(umap_transformed)
actualDistribution(umap_transformed,true_labels,true_encoded_labels)
plotAllClusterModels(umap_transformed,'audio-results4/logMelSpec_muslim_features_vowel.csv-umap')

Actual Distribution of the labels in the dataset: --> 
Counter({'a': 159, 'u': 47, 'cantdecide': 13, 'misaligned': 13, 'other': 8})
Counter({0: 159, 4: 47, 1: 13, 2: 13, 3: 8})

audio-results4/logMelSpec_muslim_features_vowel.csv-umap_kmeans_labels.npy
K-Means
predicted_labels--> Counter({3: 56, 4: 55, 0: 46, 2: 45, 1: 38})
 2D representation

 3D representation

Agglomerative
predicted_labels--> Counter({0: 159, 1: 81})
 2D representation

 3D representation

Birch
predicted_labels--> Counter({9: 22, 1: 18, 4: 18, 2: 17, 6: 17, 11: 17, 5: 16, 10: 15, 0: 14, 8: 14, 3: 13, 13: 13, 7: 12, 14: 12, 15: 10, 16: 9, 12: 3})
 2D representation

 3D representation

DBSCAN
predicted_labels--> Counter({-1: 118, 1: 23, 5: 13, 8: 12, 2: 11, 6: 11, 0: 8, 3: 8, 10: 7, 4: 7, 12: 6, 9: 6, 7: 5, 11: 5})
 2D representation

 3D representation

OPTICS
predicted_labels--> Counter({-1: 186, 1: 29, 0: 25})
 2D representation

 3D representation

MEAN-SHIFT
predicted_labels--> Counter({0: 96, 1: 88, 2: 56})
 2D representation

 3D representation

Gaussian-Mixture
predicted_labels--> Counter({0: 57, 3: 53, 4: 47, 2: 45, 1: 38})
 2D representation

 3D representation
In [ ]: