BYR Achieve · 镜像论坛

LR新手求指教

2017/12/26镜像同步0 回复

怎么输出LR逻辑回归的输入项的权值以及得到最佳权值的特征码部分代码如下 ```python # plot import matplotlib.pyplot as plt def fill_with_mean(feature, df, inttype=False): mu = np.round(df[feature].dropna().mean(), 0) return df[feature].fillna(mu).astype(int) if inttype else df[feature].fillna(mu) def grid_search_model(X, Y, model, parameters, cv): CV_model = GridSearchCV(estimator=model, param_grid=parameters, cv=cv, refit=True) CV_model.fit(X, Y) print("Best Score:", CV_model.best_score_, " / Best parameters:", CV_model.best_params_) return CV_model.best_params_, CV_model def validation_roc_model(model, parameters, Xtest, Ytest): scores = model.predict_proba(Xtest) df = pd.read_table('input.txt', ) # guess WBC feature 白细胞 using mean value df['WBC'] = fill_with_mean('WBC', df) # train Y_data = df["label"].values # data: Labels X_data = df.drop("label", axis=1).values # data: Features train_X, test_X, train_y, test_y = train_test_split(X_data, Y_data, test_size=0.2, random_state=0) cv = ShuffleSplit(n_splits=100, test_size=0.2, random_state=0) # for plotting label = [] models = [] # logistic regression param_range = np.linspace(1e-4, 1e4, 100) param_grid = {'C': param_range} C, model = grid_search_model(train_X, train_y, LogisticRegression(), param_grid, cv) models.append(model) lr = LogisticRegressionCV(Cs=10, penalty='l1', cv=5, solver='liblinear', scoring='roc_auc', refit=True) lr.fit(train_X,train_y) coef = lr.coef_.ravel() colors = ['b', 'r', 'c', 'm', 'g', 'y', 'k', 'darkorange'] names = ['LR', 'SVM', 'KNN', 'NB', 'linSVM', 'DTree', 'RF', 'adaBoost'] for i, model in enumerate(models): pred_y = model.predict_proba(test_X)[:, 1] fpr, tpr, threshold = roc_curve(test_y, pred_y, pos_label=1) auc = roc_auc_score(test_y, pred_y) label.append(names[i] + '=' + str(round(auc, 3))) plt.plot(fpr, tpr, colors[i] + '-') plt.legend(label, loc=4) plt.show() ```

订阅后，新回复会通过你的通知中心匿名送达。