Issue with Matplotlib rendering dates, image

Question

I'm having an issue with Matplotlib v 3.1.3 from conda-forge with python 3.7. I have all of the dependencies required for Matplotlib. When I enter this code, which should work. I get splatter art. It's based on this youtube tutorial: https://www.youtube.com/watch?v=LWjaAiKaf8&list=PL-osiE80TeTvipOqomVEeZ1HRrcEvtZB&index=8

import matplotlib.pyplot as plt
import pandas as pd

df_train = pd.read_csv('mydata.csv', date_parser=True)
df_train.columns = ['date', 'col1', 'col2', 'col3', 'col4', 'col5']
df_train['date'] = pd.to_datetime(df_train['date'])
df_train.set_index(['date'])

x_value = df_train['date']
y_value = df_train['col4']
plt.plot_date(x_value, y_value )
plt.gcf().autofmt_xdate()
plt.show

The rendering of the matplotlib chart based on this code looks like this:

I tried another approach using the matplotlib DateFormatter and Locator. I got something resembling a line chart underneath a child's scribbling. But it had dates:

df_train = pd.read_csv('mydata.csv', date_parser=True)
df_train.columns = ['date', 'col1', 'col2', 'col3', 'col4', 'col5']
df_train['date'] = pd.to_datetime(df_train['date'])
df_train.set_index(['date'])

    # Visualize data
x_values = df_train['date']
y_values = df_train['col4']
ax = plt.gca()
plt.figure(figsize=(16, 8))
formatter = mpl_dates.DateFormatter("%Y-%m-%d")
ax.xaxis.set_major_formatter(formatter)
locator = mpl_dates.DayLocator()
ax.xaxis.set_major_locator(locator)
plt.plot(x_values, y_values)
plt.show()

Finally, if I change the code to exclude the dates: I get a perfectly rendered chart with no dates:

import matplotlib.pyplot as plt
import pandas as pd

df_train = pd.read_csv('mydata.csv', date_parser=True)
df_train.columns = ['date', 'col1', 'col2', 'col3', 'col4', 'col5']
df_train['date'] = pd.to_datetime(df_train['date'])
df_train.set_index(['date'])

x_value = df_train['date']
y_value = df_train['col4']
plt.plot(df_train['col4']
plt.gcf().autofmt_xdate()
plt.show()

I've tried closing the plots at the end to no avail. I checked the Matplotlib docs and followed it to a 'T' including using the wheel build and creating the conda channel and installing the dependencies and setting the path and includes per the documentation. I'm at my wits end. Can someone more educated on the subject give me a hand? Thanks in advance.

GinTonic · Accepted Answer · 2020-06-01 11:40:36Z

1

It seems that the default setting in plot_date() is set to scatterplots (see (https://www.geeksforgeeks.org/matplotlib-pyplot-plot_date-in-python/) in the newer versions of matplotlib.

To achieve a continuous graph based on dates, you can define the interlining in the arguments plt.plot_date(x_value, y_value, '-').

This code works for me:

import matplotlib.pyplot as plt
import pandas as pd

df_train = pd.read_csv('test.csv', date_parser=True)
df_train.columns = ['date', 'col1', 'col2', 'col3', 'col4', 'col5', 'col6']
df_train['date'] = pd.to_datetime(df_train['date'])
df_train.set_index(['date'])

x_value = df_train['date']
y_value = df_train['col4']
plt.plot_date(x_value, y_value, '-')
plt.gcf().autofmt_xdate()
plt.show()

Output:

This functionality of not using the lineplot by default is indeed questionable given that the plot also automatically changes from scatterplot to lineplot when you just change the color: plt.plot_date(x_value, y_value, 'g').

This might just be a bug in the current versions of mpl.

edited Jun 1, 2020 at 11:40

answered Jun 1, 2020 at 11:31

GinTonic

1,0308 silver badges11 bronze badges

Sign up to request clarification or add additional context in comments.

5 Comments

brohjoe Over a year ago

Thanks Chuck. I added the '-' to the plot_date() function, now I'm getting, "ValueError: Length mismatch: Expected axis has 6 elements, new values have 7 elements"

GinTonic Over a year ago

I tested the code above with the data from here. The dataframe had 7 columns, which I used in line nr. 5. You might need to adapt this line depending on the dataset you are using.

brohjoe Over a year ago

It works perfectly on your test data, but something must be wrong with my csv because I'm still getting child scribbling over chart as in chart #2 above.

GinTonic Over a year ago

Try to sort by the date before plotting. If it doesn't help, look for outliers /wrong formatted stuff in the date column.

brohjoe Over a year ago

I had too many columns when I said, df_train.columns = [...] My data is working now.

Collectives™ on Stack Overflow

Issue with Matplotlib rendering dates, image

1 Answer 1

5 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

5 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related