From a63aafe88b55c1c500646f5c93e9ff16780d37ca Mon Sep 17 00:00:00 2001
From: Pacien TRAN-GIRARD
Date: Fri, 24 Oct 2014 02:12:51 +0200
Subject: Add some stats

---
 main.py | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/main.py b/main.py
index 8187e1f..dff07f5 100644
--- a/main.py
+++ b/main.py
@@ -50,17 +50,16 @@ ban_list = map_bans(ban_dict_list)
 ########## HISTOGRAM ##########
 
 ban_durations = [ban.get_duration() for ban in ban_list]
-(ban_durations, bins) = np.histogram(ban_durations, bins=[round(365/12*x) for x in range(1, 50+2)])
+(ban_durations_bars, bins) = np.histogram(ban_durations, bins=[round(365/12*x) for x in range(1, 50+2)])
 
 print("[INFO]", "Generating histogram")
 bar_chart = pygal.Bar(legend_at_bottom=True)
 bar_chart.title = "Active Wikipedia bans by duration (%d samples)" % SAMPLES
-bar_chart.x_labels = map(str, range(1, len(ban_durations)+1))
-bar_chart.add("Number of active bans", ban_durations)
+bar_chart.x_labels = map(str, range(1, len(ban_durations_bars)+1))
+bar_chart.add("Number of active bans", ban_durations_bars)
 bar_chart.render_to_file(HIST_FILE)
 print("[INFO]", "Histogram generation complete")
 
-
 ########## NB BAN MAP ##########
 
 def count_by_country(ban_list):
@@ -116,4 +115,10 @@ print("[INFO]", "Ban duration map generation complete")
 
 print("\nTHIS WAS A TRIUMPH!")
 print("I'M MAKING A NOTE HERE:")
-print("HUGE [SUCCESS]")
+print("HUGE [SUCCESS]\n")
+
+print("Some additional stats about ban durations:")
+print("  Mean: %.2f days" % np.mean(ban_durations))
+print("  Median: %.2f days" % np.median(ban_durations))
+print("  Deviation: %.2f" % np.std(ban_durations))
+print("  Variance: %.2f" % np.var(ban_durations))
-- 
cgit v1.2.3