sklearn 机器学习对包含文本、日期和浮动列的 Excel 工作表的行进行分类，无监督

Question

我刚刚开始学习机器学习。我正在尝试处理简单的例子。 我有一个 *.csv 文件，其中包含文本、数字和日期的列。出于学习目的，我的目标是使用 KMeans 等在无监督模式下对该文件的行进行分类。

我不知道如何组合所有变压器以将其与 KMeans 一起使用。 也许有人可以帮助我提出不同的建议。

`

import numpy as np
import pandas as pd
from sklearn.cluster import KMeans
from sklearn.preprocessing import StandardScaler
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.pipeline import make_pipeline
from sklearn.base import BaseEstimator, TransformerMixin

class DateTransformer(BaseEstimator, TransformerMixin):
    def fit(self, X, y=None):
        return self

    def transform(self, X):
       X_date_encoded = np.zeros_like(X)
       valid_dates = ~np.isnat(X).all(axis=1)
       if np.any(valid_dates):
          X_date_encoded[valid_dates] = (X[valid_dates] - X[valid_dates].min()) / (
                    X[valid_dates].max() - X[valid_dates].min())
       return X_date_encoded


def classify_excel_rows(file_path, n_clusters=3):
    """
    Classify the rows of an Excel sheet using unsupervised learning (KMeans).

    Args:
    - file_path: The path to the Excel file.
    - n_clusters: The number of clusters to form.

    Returns:
    - A DataFrame with an additional column 'Cluster' indicating the cluster each row belongs to.
   """

   # Read the Excel file
   data = pd.read_csv(file_path, sep=';', skiprows=4)

   data['Value'] = data['Value'].str.replace(',', '.').astype(float)
   data['Value'] = pd.to_numeric(data['Value'], errors='coerce')

   date_format = "%d.%m.%Y"
   data['Date_1'] = pd.to_datetime(data['Date_1'], format=date_format, errors='coerce')
   data['Date_2'] = pd.to_datetime(data['Date_2'], format=date_format, errors='coerce')

   # Identify different data types
   text_columns = data.select_dtypes(include=['object']).columns.tolist()
   numeric_columns = data.select_dtypes(include=['float64']).columns
   date_columns = data.select_dtypes(include=['datetime']).columns

   # Preprocess the data
   numeric_transformer = make_pipeline(StandardScaler())
   text_transformer = make_pipeline(TfidfVectorizer())
   date_transformer = make_pipeline(DateTransformer())

   **X = ???? # how to combine all transformers to use it with KMeans?**

   [...]

   # Perform K-Means clustering
   kmeans = KMeans(n_clusters=n_clusters, random_state=42)
   labels = kmeans.fit_predict(X)

`

我已经尝试过上面的示例代码。我不知道如何使用 KMeans 处理此类数据进行无监督学习/分类。我希望根据列值对 *.csv 文件的行进行聚类。

Answer 1

我确实找到了替代方案：

import pandas as pd
from sklearn.cluster import KMeans
from sklearn.preprocessing import StandardScaler


def classify_excel_rows(file_path, n_clusters=3):
    """
    Classify the rows of an Excel sheet using unsupervised learning (KMeans).

    Args:
    - file_path: The path to the Excel file.
    - n_clusters: The number of clusters to form.

    Returns:
    - A DataFrame with an additional column 'Cluster' indicating the cluster each row belongs to.
    """

    # Read the Excel file
    data = pd.read_csv(file_path, sep=';', skiprows=4)

    data['Value'] = data['Value'].str.replace(',', '.').astype(float)
    data['Value'] = pd.to_numeric(data['Value'], errors='coerce')

    date_format = "%d.%m.%Y"
    data['Date_1'] = pd.to_datetime(data['Date_1'], format=date_format, errors='coerce')
    data['Date_2'] = pd.to_datetime(data['Date_2'], format=date_format, errors='coerce')

    data = data.ffill()
    data = pd.get_dummies(data, drop_first=True)

    data['year'] = data['Date_1'].dt.year
    data['month'] = data['Date_1'].dt.month
    data['day'] = data['Date_1'].dt.day

    data = data.drop(['Date_1', 'Date_2'], axis=1)

    scaler = StandardScaler()
    scaled_data = scaler.fit_transform(data)

    # # Identify different data types
    # text_columns = data.select_dtypes(include=['object']).columns.tolist()
    # numeric_columns = data.select_dtypes(include=['float64']).columns
    # date_columns = data.select_dtypes(include=['datetime']).columns

    # # Perform K-Means clustering
    kmeans = KMeans(n_clusters=n_clusters, random_state=42)
    kmeans.fit(scaled_data)

    # Add the cluster labels to the original DataFrame
    data['Cluster'] = kmeans.labels_

    return data