为何bvar命令报错？贝叶斯VAR代码执行遇数据错误求助

Let's break down why you're hitting this error and how to fix it. The error message clearly states your data needs to be fully numeric and free of NAs—so let's target those two requirements with your dataset (year, quantity, price):

1. Check for non-numeric columns (most likely the culprit here) #

The bvar() function expects input data to be a numeric matrix or data frame. Your "年份" (year) column is probably not stored as a numeric type:

• It might be a character string (e.g., "2020" instead of 2020)
• Or a date/time type (like Date or POSIXct)

Fix steps: #

First, verify your column types with:

str(Data)

Option A: Convert the year column to numeric

If you want to include year as a variable in your model:

Data$年份 <- as.numeric(Data$年份)

Option B: Remove the year column (if it's just a time index)

If year is only tracking the time order (not a predictor variable), exclude it from the input data:

# Keep only quantity and price for the model
model_data <- Data[, c("数量", "价格")]
x <- bvar(model_data, lags = 1, n_draw = 1000L, n_burn = 200L, verbose = FALSE)

2. Check for missing values (NAs) #

Even if your columns are numeric, hidden NAs will trigger this error. To identify where NAs are:

# Show count of NAs per column
colSums(is.na(Data))

Fix steps: #

• Remove rows with NAs: Quick and simple if you don't lose too much data:

  Data_clean <- na.omit(Data)
  

• Impute NAs: If you need to keep all observations, use interpolation (e.g., with the zoo package) or mean/median filling:

  library(zoo)
  Data_clean <- na.approx(Data) # Linear interpolation for time series
  

Final Check #

Before running bvar() again, confirm your data meets the requirements:

# Verify all columns are numeric
all(sapply(Data_clean, is.numeric))
# Verify no NAs remain
anyNA(Data_clean)

If both return TRUE, your model should run without the data error.

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