如何基于随机值矩阵实现每列含最小阈值的Waterfall（瀑布图）

Hey there! Let's tackle this problem step by step. You've got a random matrix and want to create a waterfall chart with per-column minimum thresholds—here's a solid R-based solution using ggplot2, which is super flexible for custom visualizations.

1. 准备工作与数据预处理 #

First, we'll convert your wide-format matrix into a tidy long-format data frame, then calculate the minimum threshold for each column and compute the necessary values for the waterfall chart.

# Load required packages (install first if you haven't: install.packages("tidyverse"))
library(tidyverse)

# Your original matrix generation code
set.seed(1)
n_rows <- 6
n_cols <- 5
mat_random <- matrix(runif(n = 30, 0.04, 0.06), n_rows, n_cols)

# Convert matrix to tidy data frame and calculate thresholds
df <- mat_random %>%
  as.data.frame() %>%
  # Add stage labels for each row
  mutate(Stage = paste0("Stage_", 1:n_rows)) %>%
  # Reshape to long format (easier for grouping/plotting)
  pivot_longer(cols = -Stage, names_to = "Group", values_to = "Value") %>%
  # Group by each column (Group) and compute its minimum threshold
  group_by(Group) %>%
  mutate(
    Threshold = min(Value),
    # Calculate difference between each value and the column's threshold
    Diff_from_threshold = Value - Threshold
  )

2. 两种瀑布图实现方案 #

方案A：展示各阶段与阈值的差异 #

This chart shows how each stage's value deviates from the minimum threshold of its column, with each column (group) in its own subplot for clarity.

ggplot(df, aes(x = Stage, y = Diff_from_threshold, fill = Stage)) +
  # Bar chart for threshold differences
  geom_col(position = "identity") +
  # Add labels for the difference values
  geom_text(aes(label = round(Diff_from_threshold, 5)), vjust = -0.5, size = 3) +
  # Create a subplot for each column/group
  facet_wrap(~Group, scales = "free_y") +
  # Customize labels and theme
  labs(
    title = "Waterfall Chart: Stage vs. Column Minimum Threshold",
    y = "Difference from Column Threshold",
    x = "Stage"
  ) +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))

方案B：累计式瀑布图（从阈值开始累加） #

If you want to show cumulative changes starting from the column's minimum threshold, this version tracks the running total and marks the threshold with a dashed red line.

# Calculate cumulative values for the waterfall
df_cumulative <- df %>%
  group_by(Group) %>%
  mutate(
    # Cumulative value starting from the threshold
    Cumulative = Threshold + cumsum(Diff_from_threshold),
    # Previous stage's cumulative value (starts at threshold)
    Previous = lag(Cumulative, default = Threshold),
    # Change between current and previous stage
    Cumulative_diff = Cumulative - Previous
  )

# Plot the cumulative waterfall
ggplot(df_cumulative, aes(x = Stage, y = Cumulative_diff, fill = ifelse(Cumulative_diff > 0, "Increase", "Decrease"))) +
  geom_col(position = "identity") +
  # Label the final cumulative value for each stage
  geom_text(aes(label = round(Cumulative, 5)), vjust = ifelse(Cumulative_diff > 0, -0.5, 1.5), size = 3) +
  # Add dashed line for the column's minimum threshold
  geom_hline(aes(yintercept = Threshold), linetype = "dashed", color = "red") +
  facet_wrap(~Group, scales = "free_y") +
  labs(
    title = "Cumulative Waterfall Chart with Column Minimum Threshold",
    y = "Cumulative Value",
    x = "Stage",
    fill = "Change Direction"
  ) +
  # Custom color scheme for increases/decreases
  scale_fill_manual(values = c("Increase" = "#2ecc71", "Decrease" = "#e74c3c")) +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))

自定义调整提示 #

• If you want to use a fixed global threshold instead of per-column minima, replace Threshold = min(Value) with Threshold = [your fixed value] (e.g., Threshold = 0.04).
• Adjust the rounding precision in round() to match your needs.
• Tweak colors, labels, or themes to align with your visualization requirements.

内容的提问来源于stack exchange，提问作者Sam