结合geom_point与geom_col并分面的可视化实现技术问询

Got it, let's work through how to combine geom_col, geom_point, and faceting for your portfolio data—this should work perfectly for your 8 portfolios with 30 stocks each!

First, let's formalize your sample melted data into a tidy tibble for consistency:

library(tidyverse)

# Your sample portfolio data
portfolio_df <- tibble(
  Ticker = c("AAPL", "ABBV", "AAPL", "ABBV", "AAPL", "ABBV"),
  variable = c("Factor_Risk", "Factor_Risk", "Stock_Specific_Risk", "Stock_Specific_Risk", "Weight", "Weight"),
  value = c(4.66, 1.71, 0.21, 0.07, 4.00, 1.66),
  Portfolio = c("US", "INTL", "US", "INTL", "US", "INTL")
)

Basic Combined Visualization with Faceting #

The core approach is to layer your stacked bar chart (for risk components) and points (for weights) separately, then use faceting to split the view by each portfolio. Here's a straightforward implementation:

ggplot() +
  # Stacked bars for risk components
  geom_col(
    data = portfolio_df %>% filter(variable %in% c("Factor_Risk", "Stock_Specific_Risk")),
    aes(x = Ticker, y = value, fill = variable),
    position = "stack"
  ) +
  # Points for stock weights
  geom_point(
    data = portfolio_df %>% filter(variable == "Weight"),
    aes(x = Ticker, y = value, color = variable),
    size = 4, shape = 19
  ) +
  # Split into facets by Portfolio (free x-axis avoids crowded tickers)
  facet_wrap(~Portfolio, scales = "free_x") +
  # Custom labels and styling
  labs(
    title = "Portfolio Risk Components & Stock Weights",
    x = "Stock Ticker",
    y = "Value",
    fill = "Risk Component",
    color = "Metric"
  ) +
  theme_minimal() +
  theme(plot.title = element_text(hjust = 0.5))

Key Details Explained: #

• Layered Geoms: We filter the data for each geom separately—geom_col uses only the risk variables, while geom_point targets just the weight values. This keeps each layer focused on its specific metric.
• Facetting: facet_wrap(~Portfolio) creates a separate plot for each portfolio. scales = "free_x" ensures each facet only shows the tickers present in that portfolio, which is critical for your real-world data with 30 stocks per portfolio (no more crowded x-axes!).
• Styling: Using distinct fill (for bars) and color (for points) makes it easy to distinguish between risk components and weights.

Enhanced Version: Weights Above Total Risk #

If you want to place weight points above the total stacked risk (so they don't overlap with bars), first calculate the total risk per stock, then adjust the y-position of the points:

# Calculate total risk per stock in each portfolio
total_risk <- portfolio_df %>%
  filter(variable %in% c("Factor_Risk", "Stock_Specific_Risk")) %>%
  group_by(Portfolio, Ticker) %>%
  summarise(total_risk = sum(value), .groups = "drop")

# Merge total risk with weight data
weight_with_risk <- portfolio_df %>%
  filter(variable == "Weight") %>%
  left_join(total_risk, by = c("Portfolio", "Ticker"))

# Build the enhanced plot
ggplot() +
  geom_col(
    data = portfolio_df %>% filter(variable %in% c("Factor_Risk", "Stock_Specific_Risk")),
    aes(x = Ticker, y = value, fill = variable),
    position = "stack"
  ) +
  # Place points above total risk, add labels for clarity
  geom_point(
    data = weight_with_risk,
    aes(x = Ticker, y = total_risk + value, color = variable),
    size = 4, shape = 19
  ) +
  geom_text(
    data = weight_with_risk,
    aes(x = Ticker, y = total_risk + value, label = paste0("Wt: ", value)),
    vjust = -0.5, color = "darkred"
  ) +
  # Arrange 8 portfolios into a clean 4-column grid
  facet_wrap(~Portfolio, scales = "free_x", ncol = 4) +
  labs(
    title = "Portfolio Risk Components with Weights Above Total Risk",
    x = "Stock Ticker",
    y = "Value",
    fill = "Risk Component",
    color = "Metric"
  ) +
  theme_minimal() +
  theme(plot.title = element_text(hjust = 0.5))

This version adds context by showing weights relative to total risk, and uses ncol = 4 in facet_wrap to neatly arrange your 8 portfolios into a grid.

内容的提问来源于stack exchange，提问作者bigjdawg43