This would allow a reader of your graph to infer at a glance what proportion of contributors are one-off and what proportion are repeat, by seeing how much of the monthly curve is above the yearly one.

(defun yearly-data ()
(loop for entry in +yearly-raw-data+
for year = (parse-integer entry :end 4)
for contribs = (parse-integer entry :start 5)
collect (list year contribs)))

(defun monthly-data ()
(loop for entry in +monthly-raw-data+
for (year year-end) = (parse-integer entry :end 4)
for month = (position entry *months* :test (lambda (x y) (string= x y :start1 5 :end1 8))
for contribs = (parse-integer entry :start 9)
collect (list (+ year (/ month 12)) contribs)

(defun yearly-graph ()
(with-line-chart (600 400)
(add-series “Yearly Contributors” (yearly-data))
(add-series “Monthly Contributors” (monthly-data))
;;so the yearly totals line up, the data is offset by 1
(set-axis :x nil :draw-gridlines-p nil :data-interval 1
:label-formatter #’(lambda (y)
(princ-to-string (1- y))))
(set-axis :y nil)
(save-file “yearly.png”)))

Haven’t tried the code, or even seen if it parses, so caveat emptor.