波士顿房价数据集线性回归报错：MEDV未找到但数据集含该字段

Hey there, I get it—you’re scratching your head because you know the MEDV column exists in your dataset, but R keeps throwing that frustrating error when you run your lm() call. Let’s break down what’s going on and fix this step by step:

1. Double-check your column names (hidden characters are sneaky!) #

The error mentions CAT..MEDV, which suggests your target column might have been renamed accidentally (maybe during data import or a previous operation). First, confirm exactly what columns are in your dataset with these quick checks:

# Print all column names to spot discrepancies
colnames(data)
# Or glance at the first few rows to verify
head(data)

If you see CAT..MEDV instead of MEDV, just update your model formula to match the actual column name:

mlr1 <- lm(CAT..MEDV ~ CRIM + CHAS + RM, data = data.training)

2. Verify your training set includes the target column #

It’s rare, but sometimes sampling can go wonky (though your code looks correct at first glance). Let’s confirm MEDV made it into your training split:

# Check if MEDV is in the training set columns
"MEDV" %in% colnames(data.training)
# Inspect the training set directly
head(data.training)

If the result is FALSE, re-run your sampling code—maybe a random quirk caused the column to be excluded (unlikely, but worth ruling out).

3. Confirm your dataset loaded correctly #

If you imported the Boston Housing data from a CSV or external file, double-check your import code. For example, if you used read.csv() without header=TRUE, R might have treated your column names as the first data row. Fix that with:

# Re-import with header enabled (adjust the file path as needed)
data <- read.csv("boston_housing.csv", header = TRUE)

4. Refresh the BostonHousing dataset (if using the mlbench version) #

If you’re using the built-in BostonHousing dataset from the mlbench package, it’s possible a previous operation modified the data. Reset it with:

# Install and load mlbench if you haven't already
install.packages("mlbench")
library(mlbench)
# Reload the fresh dataset
data(BostonHousing)
data <- BostonHousing
# Re-run your full workflow
n.training <- floor(nrow(data)*0.7)
id.training <- sample(1:nrow(data), n.training)
data.training <- data[id.training,]
data.test <- data[-id.training,]
mlr1 <- lm(MEDV ~ CRIM + CHAS + RM, data = data.training)

One of these steps should get your model running smoothly!

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