I'm working on a project that retrieves data from a MySQL database and plots it into a PDF. I've done this numerous times before, but never encountered a problem such as here.
The dataframe.plot() method gives a very strange xlim compared to when I do matplotlib.pyplot(x, y).
Here's my code (simplified):
from datetime import datetime
import matplotlib.dates as mdates
import matplotlib.pyplot as plt
import pandas
# For simplicity's sake, I'll just read the CSV here
data = pandas.read_csv(r"\\earth/various/home/metservices/A03/20190429/stationscheck/1224.csv", index_col=[0, 1], parse_dates=[0])
primary_station = 1224
data_prim = data[primary_station]
data_prim = data_prim.reset_index()
data_prim = data_prim.set_index(["dtg", "id_parameter"]).unstack(level=1)[primary_station].sort_index()
ax1 = plt.subplot(2, 1, 1)
data_prim.plot(ax=ax1, lw=.75)
ax2 = plt.subplot(2, 1, 2)
for column in data_prim.columns:
ax2.plot(mdates.date2num(data_prim.index.values), data_prim[column], label=column, lw=.75)
ax2.legend()
ax2.xaxis.set_major_formatter(mdates.DateFormatter("%B"))
print(list(map(datetime.fromordinal, map(int, ax1.get_xlim()))))
print(list(map(datetime.fromordinal, map(int, ax2.get_xlim()))))
plt.show()
which results in the following image and output:
[datetime.datetime(1176, 8, 12, 0, 0), datetime.datetime(1185, 8, 25, 0, 0)]
[datetime.datetime(2018, 12, 26, 0, 0), datetime.datetime(2019, 5, 12, 0, 0)]
Here's a snippet my data via print(data.head().to_string()):
stationID 1210 1212 1218 1220 1224 1232 1321 1361
dtg id_parameter
2019-01-01 06:00:00 404 NaN NaN 14.0 16.0 4.6 26.2 13.0 13.9
405 29.1 NaN 27.8 20.1 38.8 49.9 57.6 32.1
2019-01-01 18:00:00 404 NaN NaN 30.0 36.9 3.8 27.0 13.9 21.6
2019-01-02 06:00:00 404 NaN NaN 4.4 13.2 1.5 6.4 3.6 4.8
405 2.4 NaN 34.4 50.1 5.3 33.4 17.5 26.4
2019-01-02 18:00:00 404 NaN NaN 0.6 7.7 2.3 1.3 1.8 3.2
2019-01-03 06:00:00 404 NaN NaN 2.0 6.5 1.2 2.1 2.6 1.0
405 7.8 NaN 2.7 14.2 3.5 3.4 4.4 4.2
2019-01-03 18:00:00 404 NaN NaN 0.1 1.1 0.5 1.2 0.4 0.2
2019-01-04 06:00:00 404 NaN NaN 3.8 7.6 8.5 12.5 3.4 1.9
405 13.0 NaN 3.9 8.7 9.0 13.7 3.8 2.1
2019-01-04 18:00:00 404 NaN NaN 6.4 16.6 19.5 16.1 5.0 1.5
2019-01-05 06:00:00 404 NaN NaN 2.0 2.0 10.0 0.0 5.0 0.0
405 34.2 NaN 8.3 18.2 29.3 16.1 9.7 1.5
2019-01-05 18:00:00 404 NaN NaN 1.0 0.4 6.0 2.0 1.0 0.0
2019-01-06 06:00:00 404 NaN NaN 0.6 1.5 2.0 1.8 0.3 0.0
405 9.5 NaN 1.9 1.8 7.9 3.5 1.6 0.0
2019-01-06 18:00:00 404 NaN NaN 2.0 7.2 4.4 2.1 1.3 0.0
2019-01-07 06:00:00 404 NaN NaN 0.6 2.8 1.2 0.3 0.7 0.0
405 2.2 NaN 2.6 10.0 5.6 2.4 2.0 0.0
2019-01-07 18:00:00 404 NaN NaN 0.0 1.0 3.7 1.3 1.7 1.4
2019-01-08 06:00:00 404 NaN NaN 0.5 10.6 6.6 9.3 0.0 0.8
405 8.3 NaN 0.5 11.6 10.3 10.6 1.7 2.2
2019-01-08 18:00:00 404 NaN NaN 3.2 6.6 3.2 3.6 2.2 0.5
2019-01-09 06:00:00 404 NaN NaN 4.3 3.0 4.2 2.6 0.9 0.3
As far as I know, the pandas plot methods are just wrappers around matplotlib, so why is there such a difference in the xlim? It's causing problems for my project when I use the image later on.
The reason I'm using plt.subplot(3, 1, 1) instead of something like fig, ax = plt.subplots(3, 1, 1) is because I'm also using a Cartopy GeoAxes which doesn't play nice without this construction.
