dm.fillna(0)

dm['mileage'].fillna(0)

dm.fillna(method='ffill') # 바로 앞의 유효한 값으로 채움
dm.fillna(method='bfill') # 바로 뒤의 유효한 값으로 채움

m = dm['mileage'].mean()
impute_mean = dm['mileage'].fillna(m)
impute_mean.mean()
impute_mean.std()

sns.kdeplot(df['mileage'], label='original')
sns.kdeplot(impute_mean, label='mean')
plt.legend()

from sklearn.impute import KNNImputer
imputer = KNNImputer(n_neighbors=2, weights="uniform")
X = imputer.fit_transform(dm[['mileage', 'price', 'year']])
impute_knn = X[:, 0]
impute_knn.mean()
impute_knn.std()

sns.kdeplot(dm['mileage'], label='original')
sns.kdeplot(impute_knn, label='knn')
plt.legend()

import numpy as np
from sklearn.experimental import enable_iterative_imputer
from sklearn.impute import IterativeImputer
from sklearn.neighbors import KNeighborsRegressor
imputer = IterativeImputer(sample_posterior=True, random_state=42)
X1 = imputer.fit_transform(dm[['mileage', 'price', 'year']]) # 첫번째 대체
X2 = imputer.fit_transform(dm[['mileage', 'price', 'year']]) # 두번째 대체
X = np.concatenate([X1, X2], axis=0) # 두번의 대체 결과 합치기

sns.kdeplot(dm['mileage'], label='original')
sns.kdeplot(X1[:, 0], label='imputation 1')
sns.kdeplot(X2[:, 0], label='imputation 2')
sns.kdeplot(X[:, 0], label='combined')
plt.legend()

df_filled = dm.fillna(0)

dm.fillna(0, inplace=True)