コミット: 784702bf1412c01dddf813f6609d2b2e32668b76 - myprojects-hg-reborn (hg)

コミットメタ情報

• リビジョン: 784702bf1412c01dddf813f6609d2b2e32668b76 (tree)
• 日時: 2009-09-10 02:30:12
• 作者: lorenzo
• コミッター: lorenzo

ログメッセージ

A variation of the code addendum_infectitious_power2.py. I need this just to print out some extra files I need when analyzing the data from Dublin
(always in order to disambiguate visitors and tags).

変更サマリ

• add: Python-codes/addendum_infectitious_power2_and_flux.py

差分

diff -r 03d01ed75e25 -r 784702bf1412 Python-codes/addendum_infectitious_power2_and_flux.py
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/Python-codes/addendum_infectitious_power2_and_flux.py	Wed Sep 09 17:30:12 2009 +0000
@@ -0,0 +1,317 @@
+#!/usr/bin/env python
+
+import scipy as s
+import numpy as n
+import pylab as p
+
+
+
+# def visit_duration_many_tags_new(data_filename):
+#     data_arr=p.load(data_filename)
+
+#     print "I finished reading the file"
+
+#     #convert to integer
+#     data_arr=data_arr.astype("int")
+    
+#     #keep only the relevant data (time and tag id)
+#     data_arr=data_arr[:,0:2] #Now I am left with timestamps and
+#     #tag_ids
+#     tag_id_list=s.unique1d(data_arr[:,1])
+
+#     print "the number of detected tags is, ", len(tag_id_list)
+
+#     duration_arr=s.zeros(0).astype("int")
+
+#     for i in xrange(len(tag_id_list)):
+#         tag_id=tag_id_list[i]
+#         time_sel=s.where(data_arr[:,1]==tag_id)[0]
+
+#         tag_reboot_times=data_arr[time_sel,0]
+
+#         duration_single_tag=s.diff(tag_reboot_times)
+
+#         if (tag_id==1189):
+#             p.save("1189_new_visit_durations.dat", duration_single_tag,fmt='%d')
+
+#         duration_arr=s.hstack((duration_arr,duration_single_tag))
+        
+#     return (duration_arr)
+
+def select_entrance_single_tag(entrance_data):
+    len_arr=len(entrance_data)
+
+    entrance_list=s.zeros(len_arr).astype("int")-1 #array initialization
+    ip_set=set((168624164,3232235522))
+    
+    for i in xrange(len_arr):
+        my_set=set(entrance_data[i])
+        if len(ip_set & my_set)>0:
+            entrance_list[i]=1
+
+    return entrance_list
+        
+
+
+def visit_duration_single_tag(data_arr,tag_id):
+    find_id_by_tag=s.where(data_arr[:,1]==tag_id)
+    #the following may look odd but it retrieves the 1st and
+    #7th column (counted from 1, or the 0th and 6nd counted from zero)
+    #of the matrix, that is to day the timestamp and bootcount of the
+    #sighting reports of a given tag
+    retrive_time_bootcount=data_arr[find_id_by_tag[0],:][:,0:7:6]
+    #now calculate an array giving the gaps in the bootcounts and the
+    #corresponding gaps in the timestamps
+
+    #Now I fetch the last 7 columns in my array (and the rows corresponding
+    #to station IPs)
+    single_tag_entrance_data=data_arr[find_id_by_tag[0],:][:,-8:-1]
+
+    entrance_check_single_tag=select_entrance_single_tag(single_tag_entrance_data)
+
+    bootcount_list=retrive_time_bootcount[:,1]
+
+    bootcount_unique=s.unique1d(bootcount_list)
+    time_list=retrive_time_bootcount[:,0]
+
+    duration_arr=s.zeros(len(bootcount_unique)).astype("int") -1 #it
+    #will be overwritten
+
+    flag_duration_arr=s.zeros(len(bootcount_unique)).astype("int") -1
+
+    interval_arr=s.zeros(len(bootcount_unique)-1).astype("int") -1 #it
+    #will be overwritten
+
+    single_tag_jumps=s.zeros(len(bootcount_unique)-1).astype("int") 
+    #will be partially overwritten
+    
+    single_tag_visit_begin=s.zeros(len(bootcount_unique)).astype("int") -1
+    single_tag_visit_end=s.zeros(len(bootcount_unique)).astype("int") -1
+
+
+    #i=0
+    #for my_bootcount in bootcount_unique:
+        #sel=s.where(bootcount_list==my_bootcount)
+        #duration_arr[i]=max(time_list[sel])-min(time_list[sel])
+        #i+=1
+
+
+    for i  in xrange(len(bootcount_unique)):
+        my_bootcount=bootcount_unique[i]
+        sel=s.where(bootcount_list==my_bootcount)
+
+        single_tag_visit_begin[i]=min(time_list[sel])
+        single_tag_visit_end[i]=max(time_list[sel])
+
+
+        # duration_arr[i]=max(time_list[sel])-min(time_list[sel])
+
+        duration_arr[i]=single_tag_visit_end[i]-single_tag_visit_begin[i]
+
+
+        flag_duration_arr[i]=entrance_check_single_tag[sel][0]
+        #i.e. flag duration will amount to 1 if the visit
+        #begins at the entrance and to zero otherwise 
+        
+        #if (duration_arr[i]==0):
+            #print "my_bootcount is, ", my_bootcount
+            #print "time_list[sel] is, ", time_list[sel]
+            #print "tag_id is, ", tag_id
+
+        if (i>0):
+            lower=bootcount_unique[i-1]
+            upper=bootcount_unique[i]
+            sel_lower=s.where(bootcount_list==lower)[0][-1]
+            sel_upper=s.where(bootcount_list==upper)[0][0]
+            interval_arr[i-1]=time_list[sel_upper]-time_list[sel_lower]
+            single_tag_jumps[i-1]=upper-lower #this tells me how much
+            #the bootcount is incremented when it gets updated (should
+            #be incremented by 1 only)
+
+    if (tag_id==1601):
+        p.save("1601_duration_arr.dat",duration_arr,fmt='%d'  )
+        p.save("1601_bootcount_list.dat",bootcount_list,fmt='%d'  )
+        p.save("1601_bootcount_unique.dat",bootcount_unique,fmt='%d'  )
+        p.save("1601_flag_visit.dat",flag_duration_arr,fmt='%d'  )
+        p.save("1601_flag_all.dat",entrance_check_single_tag,fmt='%d')
+
+        duration_flagged=duration_arr[s.where(flag_duration_arr<0)]
+        duration_flagged=duration_flagged/60. #convert into minutes
+
+        p.save("1601_duration_min_flagged_visits.dat",duration_flagged,fmt='%d')
+
+        tag_jumps=s.where(s.diff(bootcount_list)>1)[0]+1 #I added 1 to
+        #start counting the rows from 1 instead of zero.
+
+        p.save("1601_bootcount_jumps.dat",tag_jumps,fmt='%d')
+        p.save("1601_bootcount_jumps_at_visit_intervals.dat",single_tag_jumps,fmt='%d')
+
+
+    if (tag_id==1595):
+        p.save("1595_duration_arr.dat",duration_arr,fmt='%d'  )
+        p.save("1595_bootcount_list.dat",bootcount_list,fmt='%d'  )
+        p.save("1595_bootcount_unique.dat",bootcount_unique,fmt='%d'  )
+        p.save("1595_flag_visit.dat",flag_duration_arr,fmt='%d'  )
+        p.save("1595_flag_all.dat",entrance_check_single_tag,fmt='%d')
+
+        duration_flagged=duration_arr[s.where(flag_duration_arr<0)]
+        duration_flagged=duration_flagged/60. #convert into minutes
+
+        p.save("1595_duration_min_flagged_visits.dat",duration_flagged,fmt='%d')
+
+        tag_jumps=s.where(s.diff(bootcount_list)>1)[0]+1 #I added 1 to
+        #start counting the rows from 1 instead of zero.
+
+        p.save("1595_bootcount_jumps.dat",tag_jumps,fmt='%d')
+        p.save("1595_bootcount_jumps_at_visit_intervals.dat",single_tag_jumps,fmt='%d')
+
+
+    if (tag_id==1567):
+        p.save("1567_duration_arr.dat",duration_arr,fmt='%d'  )
+        p.save("1567_bootcount_list.dat",bootcount_list,fmt='%d'  )
+        p.save("1567_bootcount_unique.dat",bootcount_unique,fmt='%d'  )
+        p.save("1567_flag_visit.dat",flag_duration_arr,fmt='%d'  )
+        p.save("1567_flag_all.dat",entrance_check_single_tag,fmt='%d')
+
+        duration_flagged=duration_arr[s.where(flag_duration_arr<0)]
+        duration_flagged=duration_flagged/60. #convert into minutes
+        
+        p.save("1567_duration_min_flagged_visits.dat",duration_flagged,fmt='%d')
+
+        tag_jumps=s.where(s.diff(bootcount_list)>1)[0]+1 #I added 1 to
+        #start counting the rows from 1 instead of zero.
+
+        p.save("1567_bootcount_jumps.dat",tag_jumps,fmt='%d')
+        p.save("1567_bootcount_jumps_at_visit_intervals.dat",single_tag_jumps,fmt='%d')
+
+
+    #NB: the number of reboots for a tag is given by the number of unique
+    #bootcounts minus 1 !!!
+    return [duration_arr, len(bootcount_unique)-1, interval_arr \
+            ,entrance_check_single_tag, flag_duration_arr, single_tag_jumps,\
+            single_tag_visit_begin,single_tag_visit_end]
+
+
+def visit_duration_many_tags(data_arr):
+    
+    tag_id_list=s.unique1d(data_arr[:,1])
+
+    p.save("all_visit_tag_ids.dat", tag_id_list,fmt='%d' )
+    res=s.zeros(0).astype("int")
+    count_boot=s.zeros(0).astype("int")
+    interval_between_boots=s.zeros(0).astype("int")
+
+    all_tags_jumps=s.zeros(0).astype("int")
+
+    find_entrance=s.zeros(0).astype("int")
+    flag_duration_total=s.zeros(0).astype("int")
+
+    all_visit_begin=s.zeros(0).astype("int")
+
+    all_visit_end=s.zeros(0).astype("int")
+
+    all_tag_id_list_long=s.zeros(0).astype("int")
+
+    
+    for track_tag in tag_id_list:
+        duration_and_count=visit_duration_single_tag(data_arr,track_tag)
+        single_tag_duration=duration_and_count[0]
+
+        single_tag_count_boot=duration_and_count[1]
+
+        single_tag_boot_interval=duration_and_count[2]
+
+        single_tag_entrance=duration_and_count[3]
+
+        single_tag_flag_duration=duration_and_count[4]
+        single_tag_jump=duration_and_count[5]
+
+        single_visit_begin=duration_and_count[6]
+
+        single_visit_end=duration_and_count[7]
+
+        single_tag_id_long=s.ones(len(single_visit_end)).astype("int")*track_tag
+
+
+        res=s.hstack((res,single_tag_duration))
+        count_boot=s.hstack((count_boot,single_tag_count_boot))
+        interval_between_boots=s.hstack((interval_between_boots,\
+                                         single_tag_boot_interval))
+        find_entrance=s.hstack((find_entrance,single_tag_entrance))
+
+        all_tags_jumps=s.hstack((all_tags_jumps,single_tag_jump))
+        
+        all_visit_begin=s.hstack((all_visit_begin,single_visit_begin))
+
+        all_visit_end=s.hstack((all_visit_end,single_visit_end))
+
+
+        flag_duration_total=s.hstack((flag_duration_total,single_tag_flag_duration))
+
+        all_tag_id_list_long=s.hstack((all_tag_id_list_long,single_tag_id_long))
+
+    sel_duration_flagged=s.where(flag_duration_total<0)
+    extract_duration_flagged_visits=res[sel_duration_flagged]/60. #to convert into minutes
+    p.save("all_flagged_visits_duration.dat",extract_duration_flagged_visits,fmt='%d')
+    
+    return [res, count_boot,interval_between_boots, \
+            find_entrance,flag_duration_total,all_tags_jumps,all_visit_begin,\
+            all_visit_end, all_tag_id_list_long]
+
+
+
+# data_arr=p.load("infectious.log")
+
+#filename="output_long_boot_count_number.dat"
+filename="tag_and_boot_every_20_sec.dat"
+# visit_durations_from_code_by_ciro=visit_duration_many_tags(filename)
+
+# p.save("duration_new.dat",visit_durations_from_code_by_ciro, fmt='%d')
+
+data_arr=p.load(filename)
+
+print "I finished reading the file"
+
+
+
+all_duration_and_all_counts=visit_duration_many_tags(data_arr)
+
+all_durations=all_duration_and_all_counts[0]
+p.save("all_visit_duration.dat",all_durations , fmt='%d')
+
+all_counts=all_duration_and_all_counts[1]
+p.save("all_counts.dat",all_counts , fmt='%d')
+
+all_boot_intervals=all_duration_and_all_counts[2]
+p.save("all_boot_intervals.dat",all_boot_intervals , fmt='%d')
+
+all_tag_entries=all_duration_and_all_counts[3]
+p.save("all_tag_entries.dat",all_tag_entries , fmt='%d')
+
+all_tag_flag_visits=all_duration_and_all_counts[4]
+p.save("all_visit_duration_flag.dat",all_tag_flag_visits , fmt='%d')
+
+
+all_tag_jumps=all_duration_and_all_counts[5]
+p.save("all_tag_jumps_in_bootcount.dat",all_tag_jumps , fmt='%d')
+
+all_visit_start_time=all_duration_and_all_counts[6]
+p.save("all_visit_start_time.dat",all_visit_start_time , fmt='%d')
+
+
+all_visit_end_time=all_duration_and_all_counts[7]
+p.save("all_visit_end_time.dat",all_visit_end_time , fmt='%d')
+
+
+all_visit_tag_id_long=all_duration_and_all_counts[8]
+p.save("all_visit_tag_id_long.dat",all_visit_tag_id_long , fmt='%d')
+
+
+
+print "len(s.where(all_tag_flag_visits<0)[0]) is, ", len(s.where(all_tag_flag_visits<0)[0])
+
+print " len(s.where(all_visit_start_time<0)[0]) is, ", len(s.where(all_visit_start_time<0)[0])
+
+
+
+print "So far so good"