为什么我的 F1/精度/召回率输出每行仅等于 1？

Question

需要帮助找出 for 循环仅输出 1s 的原因。当我删除循环时，它工作正常，输出合理的数据，但在循环内，创建的 df 的每一行都是 1。为什么呢？

def bootstrap(x, Nboot):

    x = np.array(x)

    models = []
    resampled_stat = []
    precision = []
    recall = []
    f1 = []

    numb_rows = len(data)

    for k in range(Nboot):
        chosen_rows = np.random.choice(numb_rows, replace=True, size=numb_rows)
        bootstrap_sample = data.iloc[chosen_rows]

        resampled_stat.append(bootstrap_sample)

        X_train, X_test, Y_train, Y_test = train_test_split(x, y, test_size = 0.2, random_state = 1)

        model = tree.DecisionTreeClassifier().fit(X_train, Y_train)

        models.append(model)

        y_pred = model.predict(X_test)

        precision.append(metrics.precision_score(Y_test, y_pred, average="macro"))
        recall.append(metrics.recall_score(Y_test, y_pred, average="macro"))
        f1.append(metrics.f1_score(Y_test, y_pred, average="macro"))

    pred_df = pd.DataFrame(
        {
            "Precision": precision,
            "Recall": recall,
            "F1": f1,
            "Models": models,
        })   
    
    return pred_df

Answer 1

我已经复制了你的代码，它对我有用。因此我建议您检查输入数据。具体

• 为什么不将

  y

  传递给函数？
• 代码中

  data

  和

  bootstrap_sample

  的作用是什么？我认为这个逻辑在代码中被部分审查。它们会以某种方式影响代码的其他部分吗？
• 您的

  y

  数组有任何类吗？也许

  y

  是一个常数，因此你的预测只有 100% 准确？

from sklearn.datasets import make_classification
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier
from sklearn import metrics
import pandas as pd
import numpy as np

def bootstrap(x, y, Nboot):

    x = np.array(x)
    y = np.array(y)

    models = []
    resampled_stat = []
    precision = []
    recall = []
    f1 = []


    for k in range(Nboot):

        X_train, X_test, Y_train, Y_test = train_test_split(x, y, test_size = 0.2, random_state = 1)

        model = DecisionTreeClassifier().fit(X_train, Y_train)

        models.append(model)

        y_pred = model.predict(X_test)

        precision.append(metrics.precision_score(Y_test, y_pred, average="macro"))
        recall.append(metrics.recall_score(Y_test, y_pred, average="macro"))
        f1.append(metrics.f1_score(Y_test, y_pred, average="macro"))

    pred_df = pd.DataFrame(
        {
            "Precision": precision,
            "Recall": recall,
            "F1": f1,
            "Models": models,
        })   
    
    return pred_df


x, y = make_classification(n_samples=100, n_features=20, n_classes=2)
bootstrap(x,y, 10)