R语言中基于行计数的GP转诊数据ggplot可视化问题

Hey there! Let's work through your two ggplot visualization tasks using your GP referral data. I'll use tidyverse tools (dplyr for data wrangling + ggplot2 for plotting) since they play nicely with tibbles, and the code will scale perfectly to your full 3000+ row dataset.

First, let's make sure we have your sample data loaded (you can skip this part if your full dataset is already in your environment):

# Load sample data
sample_data <- structure(list(LAST_NAME_GP = c("NOORDHOF", "ONBEKEND", "RAHIMTOOLA", "HIEMSTRA", "VIS", "OLDENBURG", "SLACHTER", "NOORDHOF", "VOSKUILEN", "STEVENS", "COMANS", "HIJMERING", "PHILIPS", "VIS", "LOUTER"), INSTITUTION = c("OPVOEDPOLI B.V.", "PARLAN", "PARLAN", "PARLAN", "OPVOEDPOLI B.V.", "TRIVERSUM", "ALKMAARSE PSYCHOLOGENPRAKTIJK", "TRIVERSUM", "STICHTING KRAM", "TRIVERSUM", "TRIVERSUM", "TRIVERSUM", "OPVOEDPOLI B.V.", "TRIVERSUM", "ELINE BIESHEUVEL" )), row.names = c(NA, -15L), class = c("tbl_df", "tbl", "data.frame" ))

This task focuses on counting how many times each GP appears in the dataset (each row = one referral). We'll first summarize the data, then plot it as a clean bar chart.

# Load required packages
library(tidyverse)

# Step 1: Calculate total referrals per GP
gp_total_referrals <- sample_data %>%
  count(LAST_NAME_GP, name = "total_referrals") %>%
  # Sort by total referrals (descending) so busiest GPs are first
  arrange(desc(total_referrals))

# Step 2: Plot with ggplot
ggplot(gp_total_referrals, aes(x = reorder(LAST_NAME_GP, total_referrals), y = total_referrals)) +
  geom_bar(stat = "identity", fill = "#2c3e50") +
  # Flip axes to make long GP names easier to read
  coord_flip() +
  labs(
    title = "Total Referrals by GP",
    x = "GP Last Name",
    y = "Number of Referrals"
  ) +
  theme_minimal()

Key notes for your full dataset: #

• If you have hundreds of GPs, the plot might get cluttered. You can filter to show only the top N busiest GPs with slice_max(total_referrals, n = 20) after the arrange() step.
• The reorder() function ensures bars are sorted by referral count, not alphabetical order—this makes trends way easier to spot.

Here we need to count referrals per GP-institution pair. I'll show two common visualization options, plus a pro tip for handling large datasets.

Option 1: Grouped Bar Chart (per GP, split by institution) #

This lets you compare referral counts across institutions for each GP:

# Step 1: Calculate referrals per GP + institution combination
gp_institution_referrals <- sample_data %>%
  count(LAST_NAME_GP, INSTITUTION, name = "referral_count") %>%
  # Sort to keep the busiest GP-institution pairs at the top
  arrange(desc(referral_count))

# Step 2: Plot grouped bars
ggplot(gp_institution_referrals, aes(x = reorder(LAST_NAME_GP, referral_count), y = referral_count, fill = INSTITUTION)) +
  geom_bar(stat = "identity", position = "dodge") +
  coord_flip() +
  labs(
    title = "Referrals by GP and Target Institution",
    x = "GP Last Name",
    y = "Number of Referrals",
    fill = "Institution"
  ) +
  theme_minimal() +
  # Move legend to bottom to avoid cluttering the plot
  theme(legend.position = "bottom")

Option 2: Stacked Bar Chart (each bar is a GP, segments are institutions) #

This is perfect for quickly seeing the breakdown of institutions per GP:

ggplot(gp_institution_referrals, aes(x = reorder(LAST_NAME_GP, referral_count), y = referral_count, fill = INSTITUTION)) +
  geom_bar(stat = "identity") +
  coord_flip() +
  labs(
    title = "Referral Breakdown by GP and Institution",
    x = "GP Last Name",
    y = "Number of Referrals",
    fill = "Institution"
  ) +
  theme_minimal()

Pro tip for large datasets: #

If you have tons of institutions, the legend will get messy. You can use facet_wrap(~ INSTITUTION) to create separate plots for each institution, or filter to only show the top 5 most-referred institutions with slice_max(referral_count, n = 5, by = INSTITUTION).

内容的提问来源于stack exchange，提问作者Hannie