I am creating a demo using IPython notebook. I launch the notebook in the pylab inline mode, e.g. ipython notebook --pylab=inline, and what I would like to do is progressively build a plot, modifying aspects of the plot in subsequent cells, and having the chart redisplay after each modification. For instance, I would like to have consecutive cells,
CELL 1:
from pandas.io.data import DataReader
from datetime import datetime
import matplotlib.pyplot as plt
goog = DataReader("GOOG", "yahoo", datetime(2000,1,1), datetime(2012,1,1))
close_vals = goog['Close']
plot(close_vals.index, close_vals.values)
CHART DISPLAYED INLINE
CELL 2:
xlim(datetime(2009,1,1), datetime(2010,1,1))
MODIFIED CHART DISPLAYED INLINE
However, the original chart doesn't seem to make it's way into subsequent cells, and the chart displayed in CELL 2 is empty. In order to see the original plot with the modification, I have to re-issue the plot command,
CELL 2:
plot(close_vals.index, close_vals.values)
xlim(datetime(2009,1,1), datetime(2010,1,1))
This quickly gets clunky and inelegant as I add moving average trend lines and labels. Also, working from the IPython console, this method of progressively building a plot works just fine. Anyone know of a better way to create this kind of demo in the notebook? Thanks.
UPDATE:
My final code ended up looking like this.
CELL 1:
from pandas.io.data import DataReader
from datetime import datetime
import matplotlib.pyplot as plt
goog = DataReader("GOOG", "yahoo", datetime(2000,1,1), datetime(2012,1,1))
close_vals = goog['Close']
fig, ax = subplots(1,1)
ax.plot(close_vals.index, close_vals.values,label='GOOG Stock Price')
CELL 2:
ax.set_xlim(datetime(2009,1,1), datetime(2010,1,1))
fig
CELL 3:
avg_20 = [ sum(close_vals.values[i-20:i])/20.0 for i in range(20,len(close_vals))]
avg_20_times = close_vals.index[20:]
ax.plot(avg_20_times, avg_20, label='20 day trailing average')
ax.legend()
fig
After updating ax in each subsequent cell, calling fig redisplays the plot; exactly what I was looking for. Thanks!
