分布（六）利用python绘制山脊图

HsuHeinrich

发布于 2024-03-19 12:56:04

1470

发布于 2024-03-19 12:56:04

文章被收录于专栏：HsuHeinrichHsuHeinrich

分布（六）利用python绘制山脊图

山脊图（Ridgeline chart）简介

山脊图可以同时显示几个组的数值分布情况，并且可以在同一水平下，直观地对比多个分布的变化。

通过searbon绘制山脊图

# 1950～2010年西雅图的平均气温，并展示其分布

import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
sns.set_theme(style="white", rc={"axes.facecolor": (0, 0, 0, 0)})

# 导入数据
temp = pd.read_csv('https://raw.githubusercontent.com/plotly/datasets/master/2016-weather-data-seattle.csv')
temp['month'] = pd.to_datetime(temp['Date']).dt.month

# 构造月份字典
month_dict = {1: 'january',
              2: 'february',
              3: 'march',
              4: 'april',
              5: 'may',
              6: 'june',
              7: 'july',
              8: 'august',
              9: 'september',
              10: 'october',
              11: 'november',
              12: 'december'}
temp['month'] = temp['month'].map(month_dict)

# 生成月平均温度
month_mean_serie = temp.groupby('month')['Mean_TemperatureC'].mean()
temp['mean_month'] = temp['month'].map(month_mean_serie)

temp.head()

	Date	Max_TemperatureC	Mean_TemperatureC	Min_TemperatureC	month	mean_month
0	1/1/1948	10	8.0	7.0	january	4.493982
1	1/2/1948	6	4.0	3.0	january	4.493982
2	1/3/1948	7	4.0	2.0	january	4.493982
3	1/4/1948	7	4.0	2.0	january	4.493982
4	1/5/1948	7	3.0	0.0	january	4.493982

# 利用FacetGrid绘制山脊图

# 调色板
pal = sns.color_palette(palette='coolwarm', n_colors=12)

# 初始12个月份的画布
g = sns.FacetGrid(temp, row='month', hue='mean_month', aspect=15, height=0.75, palette=pal)

# 生成每个月温度Mean_TemperatureC的密度图
g.map(sns.kdeplot, 'Mean_TemperatureC',
      bw_adjust=1, clip_on=False,
      fill=True, alpha=1, linewidth=1.5)

# 给每个密度轮廓增添一条白线
g.map(sns.kdeplot, 'Mean_TemperatureC', 
      bw_adjust=1, clip_on=False, 
      color="w", lw=2)

# 给每个密度图添加水平线
g.map(plt.axhline, y=0,
      lw=2, clip_on=False)

# 增加对应的月份文本（颜色相对应）
for i, ax in enumerate(g.axes.flat):
    ax.text(-15, 0.02, month_dict[i+1],
            fontweight='bold', fontsize=15,
            color=ax.lines[-1].get_color())

# 重叠子图
g.fig.subplots_adjust(hspace=-0.3)

# 删除标题、标签、和边框
g.set_titles("")
g.set_ylabels("")
g.set(yticks=[])
g.despine(bottom=True, left=True)


# 设置标题、标签以及相关参数美化
plt.setp(ax.get_xticklabels(), fontsize=15, fontweight='bold')
plt.xlabel('Temperature in degree Celsius', fontweight='bold', fontsize=15)
g.fig.suptitle('Daily average temperature in Seattle per month',
               ha='right',
               fontsize=20,
               fontweight=20)

plt.show()

通过plotly绘制山脊图

# 1950～2010年西雅图的平均气温，并展示其分布

import plotly.graph_objects as go
import numpy as np
import pandas as pd
sns.set_theme(style="white", rc={"axes.facecolor": (0, 0, 0, 0)})

# 导入数据
temp = pd.read_csv('https://raw.githubusercontent.com/plotly/datasets/master/2016-weather-data-seattle.csv')
temp['year'] = pd.to_datetime(temp['Date']).dt.year

# 构造年份list
year_list = [1950, 1960, 1970, 1980, 1990, 2000, 2010]
temp = temp[temp['year'].isin(year_list)]

# 计算每年各平均温度数量
temp = temp.groupby(['year', 'Mean_TemperatureC']).agg({'Mean_TemperatureC': 'count'}).rename(columns={'Mean_TemperatureC': 'count'}).reset_index()

# 数据展示
temp.head()

	year	Mean_TemperatureC	count
0	1950	-12.0	1
1	1950	-11.0	1
2	1950	-10.0	2
3	1950	-9.0	6
4	1950	-8.0	4

# 将df转为字典，方便遍历
array_dict = {} # 初始空字典
for year in year_list:
    array_dict[f'x_{year}'] = temp[temp['year']==year]['Mean_TemperatureC'] # 存储每年温度
    array_dict[f'y_{year}'] = temp[temp['year']==year]['count'] # 存储每年温度计数
    array_dict[f'y_{year}'] = (array_dict[f'y_{year}'] - array_dict[f'y_{year}'].min()) \
                                / (array_dict[f'y_{year}'].max() - array_dict[f'y_{year}'].min()) # 标准化处理温度计数

fig = go.Figure()

for index, year in enumerate(year_list):
    # 绘制基础画板
    fig.add_trace(go.Scatter(
                    x=[-20, 40], y=np.full(2, len(year_list)-index),
                    mode='lines',
                    line_color='white'))
    
    # 绘制散点图
    fig.add_trace(go.Scatter(
                    x=array_dict[f'x_{year}'],
                    y=array_dict[f'y_{year}'] + (len(year_list)-index) + 0.4, # y标签增加高度，呈现堆叠状态（忽略数值意义）
                    fill='tonexty',
                    name=f'{year}'))
    
    # 添加文本信息
    fig.add_annotation(
                x=-20,
                y=len(year_list)-index,
                text=f'{year}',
                showarrow=False,
                yshift=10)
    
    # 修改标题等信息
    fig.update_layout(
                title='Average temperature from 1950 until 2010 in Seattle',
                showlegend=False,
                xaxis=dict(title='Temperature in degree Celsius'),
                yaxis=dict(showticklabels=False) # that way you hide the y axis ticks labels
                )
    
fig.show()