使用group by后，线图不是python散景中的propper

Question

在代码使用组后，线图不是python散景中的propper

newData = data.groupby([data.OrderDate.dt.year, 'Category'])['Sales'].sum().reset_index()

df = newData #pd.DataFrame.from_dict(data)
df['OrderDate'] = pd.to_datetime(df['OrderDate'])
df = df.sort_values(by='OrderDate')
categories = df['Category'].unique()    

p = figure(x_axis_type="datetime")
for cat, color in zip(categories, Spectral3):
    df_line = df.loc[df['Category'] == cat]    
    p.line(x='OrderDate', y='Sales', line_width=2, source=df_line, legend=cat, color=color)
p.legend.click_policy="hide"
show(p)

使用的数据是这个

    OrderDate         Category        Sales
0        2014        Furniture   82832.5164
1        2014  Office Supplies   86661.4220
2        2014       Technology  109617.4640
3        2015        Furniture   84801.7080
4        2015  Office Supplies   63019.8670
5        2015       Technology   82248.2500
6        2016        Furniture   85607.4175
7        2016  Office Supplies   82121.6480
8        2016       Technology   96572.8750
9        2017        Furniture  100269.3538
10       2017  Office Supplies  125699.4660
11       2017       Technology  140890.0350

但问题是线图看起来不像线图那么糟糕。

Answer 1

from bokeh.models import ColumnDataSource
from bokeh.plotting import figure
from bokeh.io import show

p = figure(x_range = (df['OrderDate'].unique().min(),df['OrderDate'].unique().max()) , y_range = (50000, df['Sales'].max() + 50000), plot_height =500, plot_width = 500)

colorlist = ['red', 'orange','green']
for cat, color in zip(df['Category'].unique(), colorlist):
    data = df[df['Category'] == cat].to_dict(orient = 'list')
    source = ColumnDataSource(data = data)
    p.line(x = 'OrderDate', y = 'Sales', source = source, color = color, line_width=2)

show(p)

这将为每个类别在同一图表中生成3行。 Bokeh喜欢将数据作为dicts传入，因此您必须创建要绘制的数据的dict，然后将其传递给ColumnDataSource 。