uczenie_maszynowe_zadania/cw_8/main.ipynb

from sklearn.cluster import KMeans
from sklearn.decomposition import PCA
from sklearn.preprocessing import StandardScaler
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

Formatowanie danych

data = pd.read_csv('flats_for_clustering.tsv',sep='\t')
data['Piętro'] = data['Piętro'].replace('parter','0')
data['Piętro'] = data['Piętro'].replace('niski parter','0')
data["Piętro"] = data["Piętro"].apply(pd.to_numeric, errors="coerce")
data = data.dropna()
scaler = StandardScaler()
scaler.fit(data)
scaled_data = scaler.transform(data)
scaled_data = pd.DataFrame(scaled_data)
scaled_data.head()

	0	1	2	3	4
0	-0.388640	-0.499154	-0.618683	-0.193055	-0.986574
1	0.481205	0.089217	-0.618683	-0.737328	-0.263285
2	0.072479	0.994405	0.503142	2.800447	0.821648
3	0.208721	0.270255	0.503142	-1.009464	-0.263285
4	0.150328	0.089217	0.503142	1.984038	3.353159

sns.pairplot(data)

<seaborn.axisgrid.PairGrid at 0x7f0d6a5819d0>

Algorytm K średnich

kmeans = KMeans(n_clusters=5,init='k-means++')
kmeans = kmeans.fit(data)

kmeans.cluster_centers_

array([[5.34232145e+05, 7.85017921e+01, 3.25448029e+00, 4.91756272e+00,
        2.63082437e+00],
       [2.30703732e+05, 4.16620939e+01, 1.98700361e+00, 5.51841155e+00,
        2.55379061e+00],
       [1.61545162e+06, 1.51384615e+02, 4.03846154e+00, 2.69230769e+00,
        1.23076923e+00],
       [3.52262093e+05, 5.68337226e+01, 2.64128912e+00, 6.15226530e+00,
        2.90565156e+00],
       [9.08821504e+05, 1.11654135e+02, 3.74436090e+00, 4.48120301e+00,
        2.38345865e+00]])

data['Clusters'] = kmeans.labels_

data.head()

	cena	Powierzchnia w m2	Liczba pokoi	Liczba pięter w budynku	Piętro	Clusters
0	290386	46	2	5.0	0.0	1
1	450000	59	2	3.0	2.0	0
2	375000	79	3	16.0	5.0	3
3	400000	63	3	2.0	2.0	3
4	389285	59	3	13.0	12.0	3

data['Clusters'].value_counts()

3    2141
1    1385
0     558
4     133
2      26
Name: Clusters, dtype: int64

sns.pairplot(data,hue='Clusters',palette='Spectral')

<seaborn.axisgrid.PairGrid at 0x7f0d7826b250>

Algorytm PCA

del data['Clusters']

pca = PCA(n_components=2)
pca.fit(scaled_data)
x_pca = pca.transform(scaled_data)
x_pca = pd.DataFrame(x_pca)
x_pca.columns = ['X','Y']
x_pca.head()

	X	Y
0	-0.733842	-0.981792
1	0.118314	-0.727320
2	0.433828	2.663883
3	0.725795	-0.774859
4	-0.195982	3.810580

sns.scatterplot(x_pca,x='X',y='Y')

<AxesSubplot:xlabel='X', ylabel='Y'>