Merging data points in NLTK's ConditionalFreqDist - python-3.x

Question

I have a Prussian newspaper corpus covering the years from 1863 to 1894 and want to plot the word usage over time. The corpus consits of roughly 2400 xml files, one file for each issue. If I would plot the ConditionalFreqDist I would get a graph with 2400 data points on the x-axis, which renders the graph unreadable.
How can I merge the information concerning the same year, displaying the average usage of each word in my search list u_input? E.g: I have 3 files for the year 1863, looking for the word 'König' - king (among other search terms), the first file contains 1 mention, the 2nd file 3 and the 3rd file 2. I would like the graph to only have one data point '1863' with the value '2'.
The plotting function:
def _plot_input():
cfd = nltk.ConditionalFreqDist(
(target, fileid[:-4]) # takes first 4 characters as lable names = year
for fileid in reader.fileids() # for all files in directory
for w in reader.words(fileid) # for all words in each file
for target in u_input
if w.lower().startswith(target) # includes words like 'könliglich' if search term was 'König'
)
cfd.plot(title='Word usage over time in Prussian Newspapers')
u_input is a list containing the words I'm analyzing, reader is my corpusreader object, files are named like this yyyy-mm-dd.xml, e.g. "1867-03-06.xml".
Thanks in advance.
Edit:
The quick fix would be to loop over all files, read all files beginning with the same year and write the contents into one single new file for each year.

Answer 1

To extract the year from the filename you must write fileid[:4], not fileid[:-4]. Once you do that, you'll have only as many x positions as there are distinct years in your corpus. This is exactly equivalent to the "quick fix" you suggest.
However, the y values will be totals for the year, not per-file averages within each year as you ask. If this is really what you needed, edit your question to clarify. (I suspect that what you really need is an average over the total number of words in a year; anything else is nonsense, unless all your files are exactly the same size.)