Seaborn, matplotlib not recognizing time as time

Question

My seaborn/matplotib plot is not recognizing time as time.

     temp         t           d
0    39.9  00:24:23  03-21-2021
1    39.9  00:28:32  03-21-2021
2    39.4  00:32:41  03-21-2021
3    39.2  00:36:48  03-21-2021
4    38.8  00:40:57  03-21-2021
..    ...       ...         ...
240  59.0  17:02:14  03-21-2021
241  58.5  17:06:23  03-21-2021
242  58.5  17:10:31  03-21-2021
243  58.5  17:14:40  03-21-2021
244  58.1  17:18:49  03-21-2021

I'm trying to maker my x-ticks round-up to the hour

my research has led me to believe my time data is not formatted as a in a way that matplotlib recognizes as time.

this is the area of my code that generates the plot:

#panda datafram is 'values'

g=sns.lineplot(x='t', y='temp', data=values,color="darkblue")


plt.ylim(values['temp'].min()-1, values['temp'].max()+1)
plt.xticks(rotation=90)
# compensate for axis labels getting clipped off
plt.subplots_adjust(bottom=0.15, left=0.15)

g.xaxis.set_major_formatter(mdates.DateFormatter('%H:%M:%S'))



g.xaxis.set_major_locator(mdates.HourLocator(interval = 400))
g.set(xlabel='Time', ylabel='Temperature')

#create unique filename based on time
filename= "Graph-" +str(dt.datetime.now().hour)+"-"+str(dt.datetime.now().minute)+".png"

plt.savefig(filename)

If its not clear look at my X axis in my plot, the hours are all messed up ( 17, 9, 1) if you look at my data frame the time data is obvious.

I can either save my time data in a different format or I need to reformat my time data for seaborn/matplotlib????

Answer 1

Since values.t is currently type object , convert it to_datetime() before calling sns.lineplot() :

values.t = pd.to_datetime(values.t)
g = sns.lineplot(x='t', y='temp', data=values, color='darkblue')

(You can verify the conversion by checking whether values.t.dtype is some kind of datetime instead of object .)

Answer 2

You can pass the name of the column containing you string-formatted timestamps to the parse_dates argument when reading your csv file. The important line would therefore be:

df = pd.read_csv(data, sep='\s+', parse_dates=['t'])

I used your data snippet to create some sample data spanning some hours and built some code based on this SO answer .

My code snippet looks like this:

import io

import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.dates as mdates


data_string = """
temp  t         d
39.9  00:24:23  03-21-2021
39.9  00:28:32  03-21-2021
39.4  00:32:41  03-21-2021
39.2  00:36:48  03-21-2021
38.8  00:40:57  03-21-2021
39.9  00:50:23  03-21-2021
39.9  01:28:32  03-21-2021
39.4  01:32:41  03-21-2021
39.2  01:36:48  03-21-2021
38.8  01:40:57  03-21-2021
39.9  01:50:23  03-21-2021
39.9  02:24:23  03-21-2021
39.9  02:28:32  03-21-2021
39.4  02:32:41  03-21-2021
39.2  02:36:48  03-21-2021
38.8  02:40:57  03-21-2021
39.9  02:50:23  03-21-2021
39.9  03:28:32  03-21-2021
39.4  03:32:41  03-21-2021
39.2  03:36:48  03-21-2021
38.8  03:40:57  03-21-2021
39.9  03:50:23  03-21-2021
39.9  04:24:23  03-21-2021
39.9  04:28:32  03-21-2021
39.4  04:32:41  03-21-2021
39.2  04:36:48  03-21-2021
38.8  04:40:57  03-21-2021
39.9  04:50:23  03-21-2021
39.9  05:28:32  03-21-2021
39.4  05:32:41  03-21-2021
39.2  05:36:48  03-21-2021
38.8  05:40:57  03-21-2021
39.9  05:50:23  03-21-2021
"""

data = io.StringIO(data_string)
df = pd.read_csv(data, sep='\s+', parse_dates=['t'])

print(df)


hours = mdates.HourLocator(interval=1)
h_fmt = mdates.DateFormatter('%H:%M:%S')

fig, ax = plt.subplots()

ax.plot(df['t'], df['temp'])

ax.xaxis.set_major_locator(hours)
ax.xaxis.set_major_formatter(h_fmt)

fig.autofmt_xdate()

plt.show()

The following plot will be generated:

Answer 3

Here is another way to do it. Combine your date and time columns and convert them to a column of datetime objects, then use matplotlib.dates.date2num() to convert the datetimes to numerical value than you can use for the x axis:

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

df = pd.read_csv('sample.csv')
df['dt_string'] = df.d + ' ' + df.t
df['dt'] = df['dt_string'].apply(lambda x: datetime.datetime.strptime(x, "%m-%d-%Y %H:%M:%S"))
df['time_num'] = mdates.date2num(df.dt)
print(df)

#the first 5 rows of your dataframe:
   temp         t  ...                  dt      time_num
0  39.9  00:24:23  ... 2021-03-21 00:24:23  18707.016933
1  39.9  00:28:32  ... 2021-03-21 00:28:32  18707.019815
2  39.4  00:32:41  ... 2021-03-21 00:32:41  18707.022697
3  39.2  00:36:48  ... 2021-03-21 00:36:48  18707.025556
4  38.8  00:40:57  ... 2021-03-21 00:40:57  18707.028438

Then plot as follows:

fig, ax = plt.subplots()
ax.plot(df.time_num, df.temp)
ax.set_xticks(df.dt)
ax.set_xticklabels(df.dt)
ax.set_xlabel('DateTime')
ax.set_ylabel('Time')
fig.autofmt_xdate()
plt.show()