如何在Python的Pandas DataFrame中获取数据列的斜率？

Got it, let's break down how to compute the slope for each of your columns (A, B, C, D) using scikit-learn's linear regression model, then wrap the results into a clean Pandas Series.

Step 1: Import Required Libraries #

First, make sure you have the necessary packages installed, then import them:

import pandas as pd
from sklearn.linear_model import LinearRegression
import numpy as np

Step 2: Prepare Your Data (Example Included) #

I'll use a sample DataFrame for demonstration—replace this with your actual data:

# Generate sample time-series-like data (cumsum of random noise)
np.random.seed(42)
df = pd.DataFrame({
    'A': np.random.randn(100).cumsum(),
    'B': np.random.randn(100).cumsum() + 2,
    'C': np.random.randn(100).cumsum() - 1,
    'D': np.random.randn(100).cumsum() + 0.5
})

Step 3: Define a Slope Calculation Function #

We'll create a helper function to compute the slope for a single column. Here, we're using the row index as the independent variable x (if you have a different x variable like time, just swap that in):

def get_column_slope(column_data):
    # Create 2D array for x (required by scikit-learn)
    x = np.arange(len(column_data)).reshape(-1, 1)
    # Reshape y to match scikit-learn's input format
    y = column_data.values.reshape(-1, 1)
    
    # Initialize and fit the linear regression model
    lr_model = LinearRegression()
    lr_model.fit(x, y)
    
    # Return the slope (coefficient for x)
    return lr_model.coef_[0][0]

Step 4: Compute Slopes for All Target Columns #

Use Pandas' apply method to run our function on each column, which will automatically return a Series:

# Calculate slopes for columns A, B, C, D
slope_results = df[['A', 'B', 'C', 'D']].apply(get_column_slope)

# Print the result (a pd.Series with column names as index)
print(slope_results)

Example Output #

You'll get something like this (values are dummy, based on the sample data):

A    0.006909
B    0.010113
C   -0.003403
D    0.008705
dtype: float64

Key Notes #

• If your independent variable x isn't the row index (e.g., a datetime column), replace np.arange(len(column_data)) with your x values (make sure to reshape to 2D with .reshape(-1,1)).
• The coef_ attribute of the fitted model gives the slope(s)—since we have a single x variable, we access coef_[0][0] to get the scalar slope value.

内容的提问来源于stack exchange，提问作者Mike