在 matplotlib.pyplot 中绘制带有时区的日期

Question

How do I properly set pyplot to plot dates in a given timezone? I can see various places to put a tzinfo object, but all of these either do nothing or do not remotely work as expected. I load a list of records from a csv, with each record being [naive timestamp in UTC, number to plot]

import pytz
from datetime import datetime
import matplotlib.pyplot as plt
import matplotlib.dates as mdates
estZone=pytz.timezone('US/Eastern')

records=[['10/10/17 12:00','100'],['11/10/17 12:00','150'],['12/10/17 12:00','200']]

def parseDT(stampString):
    return datetime.strptime(stampString,'%d/%m/%y %H:%M')

def plotGraph():
    xvalues=[]
    yvalues=[]
    for r in records:
        rdatetime=(parseDT(r[0])) #parseDT just returns naive DT
        rdatetime=pytz.utc.localize(rdatetime)
        rdatetime=rdatetime.astimezone(tz=estZone)
        xvalues.append(rdatetime)
        yvalues.append(int(r[1]))

    fig,ax=plt.subplots(figsize=(8,3))
    plt.plot_date(xvalues,yvalues,ls='solid',lw=1.5,aa=True,marker='None',color='g')
    dayFmt=mdates.DateFormatter('%a-%b-%d')
    hrFmt=mdates.DateFormatter('%H:00')
    ax.xaxis.set_major_formatter(dayFmt)
    ax.xaxis.set_major_locator(mdates.DayLocator())
    ax.xaxis.set_minor_locator(mdates.HourLocator(byhour=[6,12,18]))
    ax.xaxis.set_minor_formatter(hrFmt)
    plt.grid(b=True, which='both',linestyle='dotted')
    plt.show()

plotGraph()

This produces (minus some formatting) this figure:

Which is actually still UTC. It's identical to the result if I had omitted the astimezone(tz=estZone) , even though I've verified that part is definitely functioning.

If I proceed to pass tz=estZone to plt.plot_date , the figure is identical. There's no change. How can I change this to make an EST graph from my UTC timestamps?

Answer 1

I had tried passing the timezone to the DateFormatters for the x-axis labels before, but the timezone modifier seems to be applied after the locators are placed, leaving all the labels reading 4 hours behind each midnight or the hours specified. I've realized that the locators also need the timezone as a parameter to counteract this. That'd do it.

Answer 2

I've been going through this for the past hour as well: the source data had the correct timezone, but Matplotlib was ignoring it. The fix is passing the timezone to the locators and formatters.

Example code to clarify:

plt.gca().xaxis.set_major_locator(mdates.DayLocator(interval=1, tz=local_tz))
plt.gca().xaxis.set_major_formatter(mdates.DateFormatter('%d/%m/%y', tz=local_tz))