Plotting in matplotlib using pandas

Question

I am trying to plot with 2 y axis and a shared x axis. I have gotten to this point, but its not the x axis I want.

Here is my code:

import pandas as pd
import matplotlib.pyplot as plt

df = pd.read_excel("Live_data_test.xlsx","Sheet1")



timedat = df.loc[:, 'Time']
temperaturedat = df.loc[:, 'Temperature']
humiditydat = df.loc[:, 'Humidity']

fig, ax1 = plt.subplots()

ax2 = ax1.twinx()
ax1.plot(timedat, temperaturedat, 'g-')
ax2.plot(timedat, humiditydat, 'b-')

ax1.set_xlabel('Time')
ax1.set_ylabel('Temperature', color='g')
ax2.set_ylabel('Humidity', color='b')

plt.show()

The x axis is plotting as 0,1,2,... for however many points I have. It is not plotting the defined x axis which should be unix time that I have set in the spreadsheet.

Spreadsheet:

Time    Temperature Humidity
1513753200  54  45
1513753201  55  48
1513753202  55  50
1513753203  56  52
1513753204  56  54
1513753205  57  54
1513753206  56  54
1513753207  56  55
1513753208  56  56
1513753209  55  56
1513753210  54  52
1513753211  53  50
1513753212  52  45
1513753213  51  45

When I do print(timedat) I get this:

0     1.513753e+09
1     1.513753e+09
2     1.513753e+09
3     1.513753e+09
4     1.513753e+09
5     1.513753e+09
6     1.513753e+09
7     1.513753e+09
8     1.513753e+09
9     1.513753e+09
10    1.513753e+09
11    1.513753e+09
12    1.513753e+09
13    1.513753e+09
Name: Time, dtype: float64

I believe converting unix time to H:M:S M/D/Y time should be simple enough. I have been searching for hours trying to plot the x axis as the defined time, but to no avail.

Try converting it to string, and then plotting. df.Time = df.Time.astype(str) — cs95
– cs95, Commented Dec 20, 2017 at 20:14
I think that may have done the trick! I just need to convert unix time to datetime and angle the tick marks labels and it will be good! Thanks! — smashed_
– smashed_, Commented Dec 20, 2017 at 20:18

Peter Leimbigler · Accepted Answer · 2017-12-20 20:27:15Z

2

Convert your epoch timestamps to datetime and use pandas built-in axis label formatting. Try replacing this:

timedat = df.loc[:, 'Time']

with this:

timedat = pd.to_datetime(df['Time'], unit='s')

answered Dec 20, 2017 at 20:27

Peter Leimbigler

11.1k1 gold badge27 silver badges39 bronze badges

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

Comments

cs95 · Accepted Answer · 2017-12-21 15:53:18Z

1

To convert your Time column from unix format to string datetime, use pd.to_datetime + dt.strftime -

df.Time = pd.to_datetime(df.Time, unit='s').dt.strftime('%H:%I:%S %m/%d/%y')
df

                 Time  Temperature  Humidity
0   07:07:00 12/20/17           54        45
1   07:07:01 12/20/17           55        48
2   07:07:02 12/20/17           55        50
3   07:07:03 12/20/17           56        52
4   07:07:04 12/20/17           56        54
5   07:07:05 12/20/17           57        54
6   07:07:06 12/20/17           56        54
7   07:07:07 12/20/17           56        55
8   07:07:08 12/20/17           56        56
9   07:07:09 12/20/17           55        56
10  07:07:10 12/20/17           54        52
11  07:07:11 12/20/17           53        50
12  07:07:12 12/20/17           52        45
13  07:07:13 12/20/17           51        45

edited Dec 21, 2017 at 15:53

answered Dec 20, 2017 at 20:34

cs95

406k106 gold badges744 silver badges795 bronze badges

8 Comments

smashed_ Over a year ago

I tried this code and it worked just how i wanted it running Python 3.6 on windows 10, but now I'm in the process transferring over to my raspberry pi 3 (hoping to use this on a touchscreen eventually) and im getting an error: 'Series' object has no attribute 'dt'. If i remove everything after ".dt" in that line of code, it graphed incorrectly (how it did initially in the picture). but when I print timedat, it prints in the format: "YYYY-MM-DD H:MM:SS.ms" Could the error be because i am running the code on python 3.4.2 which is default on Pi'?

cs95 Over a year ago

@smashed_ Are you still running the same code? What is the output of pd.to_datetime(df.Time, unit='s')?

smashed_ Over a year ago

@COLDSPEED I am running the same code. I have installed the latest version of pandas onto the pi. the code only breaks with df.Time = pd.to_datetime(df.Time, unit='s').dt.strftime('%H:%I:%S %m/%d/%y') the output of pd.to_datetime(df.Time, unit='s') is: 0 2017-12-20 07:00:00 1 2017-12-20 07:00:01 2 2017-12-20 07:00:02.005000 ... but it will not plot that on the graph. It graphs as shown in the OP.

cs95 Over a year ago

@smashed_ Hmm, very puzzling. If you say that the code works on your desktop, but not on a rpi, then you could open a new question with the raspberry-pi tag and explain your problem?

cs95 Over a year ago

I am not familiar with the environment, so there's really nothing I can think of that explains this.

|

Collectives™ on Stack Overflow

Plotting in matplotlib using pandas

2 Answers 2

Comments

8 Comments

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

8 Comments

Related