Featured post

Quiz: Data PreProcessing

Monday, 22 July 2019

Gantt Chart: Horizontal bar using matplotlib for tasks with Start Time and End Time


import pandas as pd   
from datetime import datetime
import matplotlib.dates as dates
import matplotlib.pyplot as plt

def gantt_chart(df_phase):

    # Now convert them to matplotlib's internal format...

    df_phase['Start Time'] = pd.to_datetime(df_phase['Start Time'], format='%Y-%m-%d %H:%M:%S.%f')

    df_phase['End Time'] = pd.to_datetime(df_phase['End Time'], format='%Y-%m-%d %H:%M:%S.%f')

    #Convert DF columns into lists, plt will take all values in scalar or list

    sdate = df_phase['Start Time'].tolist()

    edate = df_phase['End Time'].tolist()

    plugin = df_phase['Plugin'].tolist()

    status = df_phase['Status'].tolist()

    color = []
    #Store colors for success and failure to differentiate

    for i in range(0, len(status)):

        if status[i] == 'Success':

            color.append('green')

        else:

            color.append('red')

    #Convert time to Matplotlib number format

    edate, sdate = [dates.date2num(item) for item in (edate, sdate)]

    ypos = range(len(plugin))

    fig, ax = plt.subplots()

    time_diff = edate - sdate

    # Plot the data, color is scalar or a list

    # All are in form of list

    ax.barh(ypos, time_diff, left=sdate, linewidth = 0.5, height=0.8, align='center', color=color)

    plt.yticks(ypos, plugin)

    plt.xticks(rotation=-20)

    ax.axis('tight')



    # We need to tell matplotlib that these are dates...

    ax.xaxis_date()

    plt.ylabel('Plugins', fontsize='medium', stretch = 'normal')

    # bbox_inches='tight' will prevent cutting of y-label

    fig.savefig(image_file, bbox_inches='tight')

    plt.show()





data = [['A', '2019-06-27 18:33:58.033', '2019-06-27 19:54:04.658', 'Success'], ['B', '2019-06-27 19:54:04.957', '2019-06-27 19:58:14.570', 'Success'], ['C', '2019-06-27 19:54:04.963', '2019-06-27 19:54:19.928', 'Failed']]

df_phase = pd.DataFrame(data, columns = [Plugin, 'Start Time', 'End Time', 'Status'])

gantt_chart(df_phase)