[seaborn] clustermap, heatmap으로 시각화하기

import seaborn as sns
import sklearn
from sklearn.datasets import load_diabetes , load_wine
data_dict = load_wine()
data = pd.DataFrame(data_dict.data,columns=data_dict.feature_names)

 

lut = dict(zip(data_dict.target_names, "rbg"))
change = dict(zip(np.unique(data_dict.target) , data_dict.target_names))
row_colors  = pd.Series(data_dict.target).map(change).map(lut)
change , lut

 

Correlation Clustermap

from sklearn.preprocessing import MinMaxScaler
m = MinMaxScaler(feature_range=(-1,1))
data_num_minmax = m.fit_transform(data)
data_num_minmax = pd.DataFrame(data_num_minmax , 
                               columns=data.columns.tolist())
sns.clustermap(data_num_minmax, 
               annot = False,      
               cmap = 'RdYlBu_r',  
               vmin = -1, vmax = 1, 
               figsize=(15,15),
               metric="correlation"
              )
## https://seaborn.pydata.org/examples/structured_heatmap.html

---

Correlation Clustermap + row_colors

sns.clustermap(data_num_minmax, 
               annot = False,      # 실제 값 화면에 나타내기
               cmap = 'RdYlBu_r',  # Red, Yellow, Blue 색상으로 표시
               vmin = -1, vmax = 1, #컬러차트 -1 ~ 1 범위로 표시
               figsize=(15,15),
               metric="correlation",
               row_colors=row_colors)

---

Correlation Heatmap

corr_matrix = data.corr().abs()
mask = np.zeros_like(corr_matrix, dtype=np.bool)
mask[np.triu_indices_from(mask)]= True
corr_matrix

corr_rel_check = (corr_matrix > 0.8).sum()-1 
pd.DataFrame(corr_rel_check[corr_rel_check > 0]).T

 

import seaborn as sns
f, ax = plt.subplots(figsize=(15, 15))
heatmap = sns.heatmap(corr_matrix,
                      mask = mask,
                      square = True,
                      linewidths = .5,
                      cmap = 'RdYlBu_r',
                      cbar_kws = {'shrink': .4,
                                'ticks' : [0 , 0.25, 0.5, 0.75 , 1]},
                      vmin = 0,
                      vmax = 1,
                      annot = True,
                      annot_kws = {'size': 13}
                     )
sns.set_style({'xtick.bottom': True}, {'ytick.left': True})