72 lines
2.4 KiB
Python
72 lines
2.4 KiB
Python
|
from sklearn.linear_model import LassoCV
|
|||
|
|
from sklearn.ensemble import RandomForestRegressor
|
|||
|
|
from sklearn.feature_selection import RFE
|
|||
|
|
from sklearn.preprocessing import MinMaxScaler
|
|||
|
|
import numpy as np
|
|||
|
|
import pandas as pd
|
|||
|
|
|
|||
|
|
def rank_to_dict(ranks, names):
|
|||
|
|
ranks = np.abs(ranks)
|
|||
|
|
minmax = MinMaxScaler()
|
|||
|
|
ranks = minmax.fit_transform(np.array(ranks).reshape(14, 1)).ravel()
|
|||
|
|
ranks = map(lambda x: round(x, 2), ranks)
|
|||
|
|
return dict(zip(names, ranks))
|
|||
|
|
|
|||
|
|
np.random.seed(0)
|
|||
|
|
size = 750
|
|||
|
|
X = np.random.uniform(0, 1, (size, 14))
|
|||
|
|
|
|||
|
|
Y = (10 * np.sin(np.pi * X[:, 0] * X[:, 1]) + 20 * (X[:, 2] - .5) ** 2 +
|
|||
|
|
10 * X[:, 3] + 5 * X[:, 4] ** 5 + np.random.normal(0, 1, size))
|
|||
|
|
|
|||
|
|
X[:, 10:] = X[:, :4] + np.random.normal(0, .025, (size, 4))
|
|||
|
|
|
|||
|
|
lasso_cv = LassoCV(alphas=np.linspace(0.001, 1, 100), cv=5)
|
|||
|
|
lasso_cv.fit(X, Y)
|
|||
|
|
|
|||
|
|
rf = RandomForestRegressor(n_estimators=100)
|
|||
|
|
rfe = RFE(estimator=rf, n_features_to_select=1, step=1)
|
|||
|
|
rfe.fit(X, Y)
|
|||
|
|
|
|||
|
|
# названия признаков
|
|||
|
|
names = ["x%s" % i for i in range(1, 15)]
|
|||
|
|
|
|||
|
|
# Stable Randomized Lasso Simulation
|
|||
|
|
n_resampling = 200
|
|||
|
|
rlasso_coefs = np.zeros((X.shape[1], n_resampling))
|
|||
|
|
for i in range(n_resampling):
|
|||
|
|
Y_permuted = np.random.permutation(Y)
|
|||
|
|
rlasso = LassoCV(alphas=np.linspace(0.001, 1, 100), cv=5)
|
|||
|
|
rlasso.fit(X, Y_permuted)
|
|||
|
|
rlasso_coefs[:, i] = rlasso.coef_
|
|||
|
|
|
|||
|
|
rlasso_scores = np.std(rlasso_coefs, axis=1)
|
|||
|
|
|
|||
|
|
# словарь для ранжирования
|
|||
|
|
ranks = {"Lasso": rank_to_dict(lasso_cv.coef_, names),
|
|||
|
|
"RFE": rank_to_dict(rfe.ranking_, names),
|
|||
|
|
"RandomizedLassoSim": rank_to_dict(rlasso_scores, names)}
|
|||
|
|
|
|||
|
|
mean = {}
|
|||
|
|
for method, values in ranks.items():
|
|||
|
|
for feature, score in values.items():
|
|||
|
|
# Если элемента с текущим ключом в mean нет - добавляем
|
|||
|
|
if feature not in mean:
|
|||
|
|
mean[feature] = 0
|
|||
|
|
# Суммируем значения по каждому ключу-имени признака
|
|||
|
|
mean[feature] += score
|
|||
|
|
df_ranks = pd.DataFrame(ranks)
|
|||
|
|
|
|||
|
|
# Выводим ранжирование
|
|||
|
|
print("ПО КАЖДОМУ МЕТОДУ:")
|
|||
|
|
print(df_ranks)
|
|||
|
|
|
|||
|
|
# Находим среднее по каждому признаку
|
|||
|
|
for feature, score in mean.items():
|
|||
|
|
mean[feature] = round(score / len(ranks), 2)
|
|||
|
|
|
|||
|
|
# Отсортированные средние значени
|
|||
|
|
mean = sorted(mean.items(), key=lambda x: x[1], reverse=True)
|
|||
|
|
print("СРЕДНИЕ")
|
|||
|
|
print(mean)
|