绘图不适用于周级别，仅适用于日级别

Question

我有以下数据框：

Date  rank_.            category  genre breakdown  count
0     2022-01-01        top 50    dutch hiphop      7
1     2022-01-01        top 50       dutch pop     12
2     2022-01-01        top 50           other     31
3     2022-01-01        top 50     other dutch      0
4     2022-01-01       top 100    dutch hiphop      2
...          ...           ...             ...    ...
5515  2023-04-05       top 100     other dutch      0
5516  2023-04-05       top 200    dutch hiphop     11
5517  2023-04-05       top 200       dutch pop     16
5518  2023-04-05       top 200           other     74
5519  2023-04-05       top 200     other dutch      1

如果我创建一个如下图所示的堆叠条形图，这会很好用。但是......如果我将数据框更改为具有一周格式（如

37-2022

）并尝试在x轴上绘制它我得到一个空图。

我认为它与日期格式与字符串有关......但我被卡住了。

任何帮助将不胜感激！

counts = df.groupby(['Date', 'genre breakdown'])['track_isrc'].nunique()

# create a new dataframe with the counts
counts_df = counts.reset_index(name='count')

# Calculate percentage of total views for each genre breakdown within each date
counts_df['percent'] = counts_df.groupby('Date')['count'].apply(lambda x: x / x.sum() * 100)

# Define the category order for the 'genre breakdown' variable
category_order = ['Dutch Pop', 'Dutch HipHop']

# Create a stacked bar graph with percentages
fig = px.bar(counts_df, x='Date', y='percent', color='genre breakdown', barmode='stack', 
             category_orders={"genre breakdown": category_order},
             color_discrete_sequence=px.colors.qualitative.Pastel)

Answer 1

由于像

{week number}-{year}

这样的日期格式不是很标准，如果您修改 DataFrame 中的

Date

列以适应这种格式，plotly 可能无法正确解释它。

你应该做的是直接在情节中指定

tickformat

：

fig.update_xaxes(rangeslider_visible=True, tickformat='%W-%Y')

这是一个使用类似于您的 DataFrame 的示例

counts_df

：

import numpy as np
import pandas as pd
import plotly.express as px

## create sample counts df similar to yours
np.random.seed(42)

n_days = len(pd.date_range('2022-01-01','2023-04-05'))
counts_df = pd.DataFrame({
    'Date': list(pd.date_range('2022-01-01','2023-04-05'))*4,
    'genre breakdown': np.repeat(['dutch hiphop','dutch pop','other','other dutch'], n_days),
    'count': np.concatenate([np.random.uniform(10,20,n_days), np.random.uniform(10,20,n_days),
    np.random.uniform(40,50,n_days), np.random.uniform(1,5,n_days)])
})

# Calculate percentage of total views for each genre breakdown within each date
counts_df['percent'] = counts_df.groupby('Date', group_keys=False)['count'].apply(lambda x: x / x.sum() * 100)

# Define the category order for the 'genre breakdown' variable
category_order = ['Dutch Pop', 'Dutch HipHop']

# Create a stacked bar graph with percentages
fig = px.bar(counts_df, x='Date', y='percent', color='genre breakdown', barmode='stack', 
             category_orders={"genre breakdown": category_order},
             color_discrete_sequence=px.colors.qualitative.Pastel)


fig.update_xaxes(rangeslider_visible=True, tickformat='%W-%Y')
fig.update_yaxes(range=[0,100])
fig.show()