通过 python 创建多类别图表

Question

I have the data like the following in excel:

company	month-year	#people got	interviewed	# people employed

link to the data: ( https://docs.google.com/spreadsheets/d/1DwZt9fpnzR9yUNBMjmqA1hg11d-2dXNs/edit?usp=share_link&ouid=113997824301423906122&rtpof=true&sd=true )

when I try to create multicategory chart(company as first category and the year-month as second category) by plotly library by python it mixes up the order of second category for y,z company. Putting the code and the screenshot of the chart below.

Code:

import pandas as pd
from helper_functions import get_df
import plotly.graph_objects as go
from datetime import datetime

def multicat_chart(infile=None, sheet_name=None, chart_type = None, chart_title = None):
    
    #chart type must be given
    df=pd.read_excel(infile,sheet_name)
    df = df.fillna(method='ffill')
    cat = df.columns[0]
    sub_cat = df.columns[1]
    cols = df.columns[2:]
    fig = go.Figure()
    cats = []
    sub_cats = []
    
    for c in df[cat].unique():
        new_df = df.loc[df[cat] == c]
        scats = new_df[sub_cat]
        scats = scats.apply(lambda date: datetime.strptime(date, "%b-%Y"))
        scats = list(scats)
        scats.sort()

        scats = [datetime.strftime(element, '%b-%y') for element in scats]
        scats = [str(element) for element in scats]
        for sc in scats:
            cats.append(str(c))
            sub_cats.append(str(sc))
        print(c)
        for i in scats:
            print(i)

    fig.add_trace( go.Bar(x = [cats,sub_cats],y = df[cols[0]], name="# people got interviewed" ))
    fig.add_trace( go.Bar(x = [cats,sub_cats],y = df[cols[1]], name="# people employed" ))
    fig.update_layout(width = 1000, height = 1000)
    return fig
    

fig = multicat_chart(infile = 'data_for_test.xlsx', sheet_name = 'data', chart_type = 'bar')
fig.show()

Chart:

I gave the data to the Bar() function in ordered way but it mixes somehow, I would like to have in ascending order, what I did I convert string to datetime object and then sorted all subcategory data with the sort() function of list, and converted back to string. And By running the script you can notice that it prints in the right order, it means that it is given ordered to function, but it mixes, who can help me to understand why it behaves so?

Answer 1

Once you made the dates month-year, they were object type--character strings, not dates—as in date-type. When you sorted, you sorted by calendar month.

First, use strptime to make it a date, sort it, then use strftime .

import pandas as pd
import plotly.graph_objects as go
from datetime import datetime as dt

def multicat_chart(infile=None, sheet_name=None, chart_type = None, chart_title = None):
    
    #chart type must be given
    df = pd.read_excel(infile, sheet_name)
    df = df.fillna(method='ffill')       # fill in companies

    df['month'] = [dt.strptime(x, '%b-%Y') for x in df['month']] # date for ordering
    df.sort_values(by = ['month', 'company'], inplace = True)    # appearance order
    df['month2'] = [dt.strftime(x, '%b-%Y') for x in df['month']] # visual appearance
    fig = go.Figure()  # plot it
    fig.add_trace( go.Bar(x = [df.iloc[:, 0], df.iloc[:, 4]], 
                          y = df.iloc[:, 2], name="# people got interviewed" ))
    fig.add_trace( go.Bar(x = [df.iloc[:, 0], df.iloc[:, 4]], 
                          y = df.iloc[:, 3], name="# people employed" ))
    fig.update_layout(width = 1000, height = 1000)
    return fig

You had a lot of extra work going on in your function; I cut a lot of that out because you didn't need it.

If you have any questions, let me know!