数据挖掘实践

import xgboost as xgb
from sklearn.metrics import confusion_matrix

clf = xgb.XGBClassifier(n_jobs=-1, max_depth=6, n_estimators=30,
                        learning_rate=0.01, colsample_bytree=.9, gamma=1,
                        reg_alpha=4, objective='binary:logistic',
                        subsample=0.8)   # eta 即 learning_rate，二者只传其一
clf.fit(X_train, y_train)
# pickle 保存模型，confusion_matrix 评估

from gensim.models import Word2Vec
from sklearn.cluster import KMeans
import joblib

# gensim 4.x：size 改名 vector_size，iter 改名 epochs
model = Word2Vec(getSentence(), vector_size=200, window=15, min_count=10, workers=4)
# 城市词库 mddwords.txt 匹配取词向量
wordvector = [model.wv[city] for city in cities if city in model.wv]
labels = KMeans(n_clusters=100, max_iter=100).fit_predict(wordvector)   # n_jobs 已在 sklearn 0.23 移除
joblib.dump(model, 'word2vec.model')   # 保存模型

from sklearn.linear_model import LinearRegression
from sklearn import metrics
from sklearn.model_selection import train_test_split
import numpy as np

from sklearn.datasets import fetch_california_housing
housing = fetch_california_housing()   # 20640 条、8 特征，目标为街区房屋价值中位数
X_train, X_test, y_train, y_test = train_test_split(housing.data, housing.target,
                                                    test_size=0.2, random_state=0)
reg = LinearRegression()
reg.fit(X_train, y_train)
print(reg.intercept_, reg.coef_)   # 截距与特征系数

y_pred = reg.predict(X_test)
mae = metrics.mean_absolute_error(y_test, y_pred)
mse = metrics.mean_squared_error(y_test, y_pred)
rmse = np.sqrt(metrics.mean_squared_error(y_test, y_pred))

# 取词向量
./fasttext skipgram -input data.txt -output model
# 分类（标签加 __label__ 前缀、空格分隔）
./fasttext supervised -input train.txt -output model
# 追加查询
./fasttext print-word-vectors model.bin < queries.txt
# 产出 model.bin（参数/预测）与 model.vec（词向量）