如何用DataFrame或矩阵创建热力图？R语言技术问询

Hey there! Great question—you don’t actually need to force-convert your DataFrame to a matrix to make a heatmap in R. Most popular heatmap packages work perfectly with DataFrames, as long as your data structure fits what the heatmap expects. Let’s break this down with your sample data.

First, a quick fix for your dataset: your reporting_amount column is stored as character strings, which will cause issues for plotting. Convert it to numeric first:

product <- c('Credit')
startdate <- c('12/30/2018','12/30/2018','12/30/2018','12/30/2018','12/30/2018')
reporting_amount <- c('29918501.83','50000000','40000000','13766666.67','75000000')
mydata <- data.frame(product, startdate, reporting_amount)

# Convert amount to numeric
mydata$reporting_amount <- as.numeric(mydata$reporting_amount)

Note: Your current sample has only one unique product and one unique startdate, so the heatmap will just be a single cell. I’ll assume your actual data has more variation—here’s how to handle both common heatmap workflows:

1. Using ggplot2 (works directly with long-format DataFrames) #

ggplot2 is the most flexible option, and it’s built to work with long-format DataFrames (the structure your sample uses: rows = observations, columns = variables). Even with your current data, you can plot it (though it’ll be simple):

library(ggplot2)

ggplot(mydata, aes(x = product, y = startdate, fill = reporting_amount)) +
  geom_tile(color = "white") +  # Adds borders between tiles
  scale_fill_viridis_c(option = "plasma") +  # Nice color scale
  labs(x = "Product", y = "Start Date", fill = "Reporting Amount") +
  theme_minimal()

If you add more products or dates to your data, this will automatically expand into a proper heatmap—no matrix conversion needed.

2. Using pheatmap (supports wide-format DataFrames or matrices) #

pheatmap is a popular package for more traditional heatmaps (like those used in genomics). It prefers wide-format data (rows = groups, columns = categories, cells = values). You don’t need to convert to a matrix—just pass a wide-format DataFrame:

library(pheatmap)
library(tidyr)

# Convert your long-format data to wide-format (for demonstration)
# This makes sense if you have multiple products/dates
wide_mydata <- mydata %>%
  pivot_wider(names_from = product, values_from = reporting_amount)

# Plot directly with the DataFrame
pheatmap(wide_mydata,
         main = "Reporting Amount Heatmap",
         color = viridis::viridis(100))

pheatmap will handle the DataFrame internally—no need to call as.matrix() unless you specifically want to.

Key Takeaways #

• ggplot2: Use directly with long-format DataFrames (no conversion needed, most flexible).
• pheatmap/other matrix-focused packages: Use wide-format DataFrames (convert from long if needed) — no manual matrix conversion required.
• Always ensure your numeric values are stored as numeric types, not characters!

内容的提问来源于stack exchange，提问作者ASH