convert time series plot using matplot to plotly

Question

I recently come across plotly which is a very powerful interactive online plot system. So i was thinking converting plots generated by matplot into plotly. Everything works fine except for the time series plot.

here is a simple version of my code.

import pandas as pd
import matplotlib.pyplot as plt
import plotly.plotly as py
import plotly.tools as tls
tls.set_credentials_file(
        username="edwinwang1988", 
        api_key="o8xw6s61dn")
import numpy as np

x = pd.date_range('1/1/2001',periods =72,freq='D')  # genrate a date range
Series = pd.DataFrame(x,columns=['Date']) # generate a date range data frame

np.random.seed(0)
col = ['A','B','C','D'] #set col names for return series
s = pd.DataFrame(np.random.randn(72,4),columns=col) # generate random return series with col names as col
for i in range(len(col)):
    s[col[i]] = s[col[i]].cumsum()

s['Date']=x  # add date to dataframe s
s.plot(x='Date') # plot s and set date as x axis

this works fine and give me the plot i want with date as x axis. but when i tried to convert the figure to plotly.

mpl_fig2= plt.gcf()
py.iplot_mpl(mpl_fig2,strip_styple = True,filename='test')

i am seeing code error like "min() arg is an empty sequence" and sometimes this "year=1 is before 1900; the datetime strftime() methods require year >= 1900"

theengineear · Accepted Answer · 2014-11-11 08:31:05Z

1

Full disclosure, I work for Plotly.

tl;dr:

run this in your terminal and it should be fixed:

pip install --upgrade plotly

Context:

So, pandas can do some funny things to dates in matplotlib. This is how matplotlib handles dates, basically:

day 1 of year 1 in utc is the floating point 1.0

http://matplotlib.org/api/dates_api.html#module-matplotlib.dates

Depending on how you use pandas to format your dates, you can end up with a pandas TimeSeries_DateFormatter. To the best of my knowledge, the repercussion of this is actually to change the underlying data associated with the date times to the following convention:

day 1 of year 1970 in utc is the floating point 0.0

So, days (not seconds!) since the epoch (http://en.wikipedia.org/wiki/Unix_time). Fyi, this goes back to ~9-22-1677 by using negative numbers.

Issue:

plotly was treating datetimes inside matplotlib as matplotlib datetimes, but pandas adds a formatter that changes this. A patch is added in plotly 1.4.2

answered Nov 11, 2014 at 8:31

theengineear

9577 silver badges7 bronze badges

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

1 Comment

Edwin Over a year ago

still doesn't work for some reason. I am getting the same error code. @theengineear

Collectives™ on Stack Overflow

convert time series plot using matplot to plotly

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related