如何用gganimate结合geom_line实现折线向指定折线收敛的动画？

原始代码与需求 #

用户提供的原始折线图代码如下：

library(ggplot2)
time_points <- rep(1:10, times = 5)
class <- rep(letters[1:5], each = 10)

# 设置不同类别均值
class_means <- c(0, 5, 10, 20, 30)
outcome_values <- rnorm(50, mean = rep(class_means, each = 10), sd = 1)

# 创建数据框
df <- data.frame(
  time = time_points,
  class = class,
  outcome = outcome_values
)

ggplot(df) +
  geom_line(aes(x = time, colour = class, y = outcome))

需求：

• 让a、b、d、e类的折线向c类折线"滑动"收敛，到达后其他折线消失
• 或者实现向均值折线收敛的效果
• 确认是否可用gganimate实现

方案1：向c类折线收敛并隐藏其他折线 #

完全可以用gganimate实现，核心思路是生成过渡帧数据，让非c类的outcome值逐步逼近c类对应时间点的值，再通过动画状态控制显示逻辑。

完整代码 #

library(ggplot2)
library(gganimate)

# 原始数据生成
time_points <- rep(1:10, times = 5)
class <- rep(letters[1:5], each = 10)
class_means <- c(0, 5, 10, 20, 30)
outcome_values <- rnorm(50, mean = rep(class_means, each = 10), sd = 1)
df <- data.frame(
  time = time_points,
  class = class,
  outcome = outcome_values
)

# 提取c类的基准值
c_vals <- df[df$class == "c", "outcome"]
names(c_vals) <- df[df$class == "c", "time"]

# 生成过渡数据：为每个非c类创建从原始值到c类值的渐变
n_frames <- 20  # 过渡帧数
transition_df <- lapply(unique(df$class), function(cls) {
  if (cls == "c") {
    # c类保持不变，所有帧都显示
    data.frame(
      time = df$time[df$class == cls],
      class = cls,
      outcome = df$outcome[df$class == cls],
      frame = rep(1:n_frames, each = length(df$time[df$class == cls]))
    )
  } else {
    # 非c类生成渐变值，最后一帧后隐藏
    cls_data <- df[df$class == cls, ]
    lapply(1:n_frames, function(f) {
      alpha <- f / n_frames
      new_outcome <- cls_data$outcome * (1 - alpha) + c_vals[as.character(cls_data$time)] * alpha
      data.frame(
        time = cls_data$time,
        class = cls,
        outcome = new_outcome,
        frame = f
      )
    }) %>% do.call(rbind, .) %>%
      # 最后一帧将非c类设为NA实现消失
      rbind(data.frame(
        time = cls_data$time,
        class = cls,
        outcome = NA,
        frame = n_frames + 1
      ))
  }
}) %>% do.call(rbind, .)

# 绘制动画
ggplot(transition_df, aes(x = time, y = outcome, colour = class)) +
  geom_line() +
  transition_states(frame, transition_length = 1, state_length = 0.5) +
  shadow_mark(alpha = 0.3, size = 0.5) +  # 保留运动轨迹阴影
  enter_fade() + exit_fade() +  # 平滑消失效果
  labs(title = "折线向c类收敛")

关键说明 #

• 手动构建包含过渡帧的transition_df，控制非c类值每帧逼近c类值
• transition_states负责动画状态切换，shadow_mark增强轨迹的视觉连贯性
• 最后一帧将非c类值设为NA，配合exit_fade()实现自然消失

方案2：向均值折线收敛 #

如果要向所有数据的全局均值折线收敛，只需调整基准值为对应时间点的全局均值即可，核心逻辑类似：

完整代码 #

library(ggplot2)
library(gganimate)
library(dplyr)

# 原始数据生成（同前）
time_points <- rep(1:10, times = 5)
class <- rep(letters[1:5], each = 10)
class_means <- c(0, 5, 10, 20, 30)
outcome_values <- rnorm(50, mean = rep(class_means, each = 10), sd = 1)
df <- data.frame(
  time = time_points,
  class = class,
  outcome = outcome_values
)

# 计算每个时间点的全局均值
mean_vals <- df %>%
  group_by(time) %>%
  summarise(mean_outcome = mean(outcome), .groups = "drop")
names(mean_vals$mean_outcome) <- mean_vals$time

# 生成过渡数据
n_frames <- 20
transition_df <- lapply(unique(df$class), function(cls) {
  cls_data <- df[df$class == cls, ]
  lapply(1:n_frames, function(f) {
    alpha <- f / n_frames
    new_outcome <- cls_data$outcome * (1 - alpha) + mean_vals$mean_outcome[as.character(cls_data$time)] * alpha
    data.frame(
      time = cls_data$time,
      class = cls,
      outcome = new_outcome,
      frame = f
    )
  }) %>% do.call(rbind, .) %>%
    # 最后一帧隐藏其他类，保留均值折线
    rbind(data.frame(
      time = cls_data$time,
      class = cls,
      outcome = NA,
      frame = n_frames + 1
    ))
}) %>% do.call(rbind, .)

# 添加均值折线的基础数据（确保全程显示）
mean_line_df <- mean_vals %>%
  mutate(class = "全局均值", frame = rep(1:(n_frames + 1), each = nrow(mean_vals))) %>%
  rename(outcome = mean_outcome)
transition_df <- rbind(transition_df, mean_line_df)

# 绘制动画
ggplot(transition_df, aes(x = time, y = outcome, colour = class)) +
  geom_line() +
  transition_states(frame, transition_length = 1, state_length = 0.5) +
  shadow_mark(alpha = 0.3, size = 0.5) +
  enter_fade() + exit_fade() +
  labs(title = "折线向全局均值收敛")

关键说明 #

• 用dplyr按时间分组计算全局均值，作为收敛目标
• 单独添加均值折线的全帧数据，确保动画全程可追踪目标线
• 最后一帧将非均值类设为NA，实现收敛后仅保留均值线的效果

内容的提问来源于stack exchange，提问作者Cpo