代码报错‘could not convert string to float: 'age'’的原因排查求助

Hey there, let's break down what's causing this error and fix it step by step.

The Root Cause #

That error pops up because when you use np.loadtxt("car.csv", delimiter=",") to load your data, it's trying to read the header row (the first line with column names like 'age') as numerical data. Since strings can't be converted to floats, Python throws that error. Also, your code has redundant work—you're loading the CSV once with Pandas and again with NumPy, which isn't necessary and adds confusion.

Solutions #

Let's go over two ways to fix this, with the second being the more robust option for CSV data:

1. Skip the Header with NumPy

If you want to stick with np.loadtxt, just add the skiprows=1 parameter to skip the first header line:

dataset = np.loadtxt("car.csv", delimiter=",", skiprows=1)

This tells NumPy to ignore the first line and start reading from the actual numerical data rows.

2. Use Pandas for Cleaner Data Handling (Recommended)

Since you already imported Pandas, let's lean into it—it's designed for tabular data like CSV files and handles headers automatically. Here's your revised, streamlined code:

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.model_selection import cross_val_score
from sklearn.model_selection import KFold
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import MinMaxScaler
from tensorflow.python.keras.models import Sequential
from tensorflow.python.keras.layers import Dense
from tensorflow.python.keras.wrappers.scikit_learn import KerasRegressor
import os

# Load CSV with Pandas (automatically recognizes headers)
df = pd.read_csv('car.csv')
# Replace commas with dots—make sure to assign back to df!
df = df.replace(",", ".", regex=True)
# Extract features (first 5 columns) and target (6th column)
x = df.iloc[:, 0:5].values
y = df.iloc[:, 5].values.reshape(-1, 1)

# Use separate scalers for features and target to avoid scaling conflicts
x_scaler = MinMaxScaler()
xscale = x_scaler.fit_transform(x)

y_scaler = MinMaxScaler()
yscale = y_scaler.fit_transform(y)

print("Scaling completed successfully!")

A couple key fixes here:

• df.replace(",", ".") needs to be assigned back to df—Pandas methods don't modify the original DataFrame by default.
• Using separate MinMaxScaler instances for x and y ensures their scaling rules don't interfere with each other, which will make inverse scaling (if you need to convert predictions back to original values) much easier.

Quick Recap #

The main issue was not accounting for the CSV header when loading data. Pandas simplifies this process significantly, so it's the better choice for most tabular data tasks.

内容的提问来源于stack exchange，提问作者Kalina Scarbrough