#46 Team onebillion: Fix incorrect serial numbers in tablet tracker

- Calculate percentage match between serial numbers in tracker and usage data.

team-ONEBILLION/tablet-tracker/extract_serial_numbers_not_found_in_tablet_tracker.py

@@ -45,6 +45,23 @@
     logging.debug("len(serial_numbers_in_tablet_tracker): {}".format(len(serial_numbers_in_tablet_tracker)))
 
 
+# Returns a percentage indicating how similar two tablet serial numbers are.
+# Example: (5A23002711, 5A23002751) --> 0.8
+def get_serial_match_ratio(serial1, serial2):
+    logging.debug("get_serial_match_ratio")
+
+    # Calculate number of matches
+    serial_match_count = 0
+    for index in range(0, 9):
+        if serial1[index] == serial2[index]:
+            serial_match_count += 1
+
+    serial_match_ratio = serial_match_count / 10
+    logging.debug("sequence_match({0}, {1}): {2}".format(serial1, serial2, serial_match_ratio))
+
+    return serial_match_ratio
+
+
 # Iterate the serial numbers in `tablets-uploading-data-ONEBILLION.csv`
 with open('../tablets-uploading-data/tablets-uploading-data-ONEBILLION.csv') as in_file:
     serial_numbers_not_found_in_tablet_tracker = []
@@ -84,11 +101,20 @@
 
     # Write results to a CSV file
     csv_filename = "serial-numbers-not-found-in-tablet-tracker.csv"
-    print("Writing list of missing tablet serials to the file \"" + csv_filename + "\"")
+    logging.debug("Writing list of missing tablet serials to the file \"" + csv_filename + "\"")
     with open(csv_filename, mode='w') as csv_file:
-        csv_fieldnames = ['serial_number']
+        csv_fieldnames = ['serial_number', 'closest_match', 'closest_match_percentage']
         csv_writer = csv.writer(csv_file, csv_fieldnames)
         csv_writer.writerow(csv_fieldnames)
         for serial_number in serial_numbers_not_found_in_tablet_tracker:
-            serial_number_as_list = [serial_number]
-            csv_writer.writerow(serial_number_as_list)
+            # Find the closest match in the tablet tracker
+            closest_match = None
+            closest_match_percentage = None
+            for serial_number_in_tablet_tracker in serial_numbers_in_tablet_tracker:
+                serial_match_ratio = get_serial_match_ratio(serial_number, serial_number_in_tablet_tracker)
+                if (closest_match_percentage is None) or (closest_match_percentage < serial_match_ratio):
+                    closest_match_percentage = serial_match_ratio
+                    closest_match = serial_number_in_tablet_tracker
+
+            csv_row = [serial_number, closest_match, closest_match_percentage]
+            csv_writer.writerow(csv_row)

team-ONEBILLION/tablet-tracker/extract_serial_numbers_not_found_in_tablet_usage_data.py

@@ -45,6 +45,23 @@
     logging.debug("len(serial_numbers_in_tablet_usage_data): {}".format(len(serial_numbers_in_tablet_usage_data)))
 
 
+# Returns a percentage indicating how similar two tablet serial numbers are.
+# Example: (5A23002711, 5A23002751) --> 0.8
+def get_serial_match_ratio(serial1, serial2):
+    logging.debug("get_serial_match_ratio")
+
+    # Calculate number of matches
+    serial_match_count = 0
+    for index in range(0, 9):
+        if serial1[index] == serial2[index]:
+            serial_match_count += 1
+
+    serial_match_ratio = serial_match_count / 10
+    logging.debug("sequence_match({0}, {1}): {2}".format(serial1, serial2, serial_match_ratio))
+
+    return serial_match_ratio
+
+
 # Iterate the serial numbers in the tablet tracker
 with open('tablet-tracker-ONEBILLION.csv') as in_file:
     serial_numbers_not_found_in_tablet_usage_data = []
@@ -83,11 +100,20 @@
 
     # Write results to a CSV file
     csv_filename = "serial-numbers-not-found-in-tablet-usage-data.csv"
-    print("Writing list of missing tablet serials to the file \"" + csv_filename + "\"")
+    logging.debug("Writing list of missing tablet serials to the file \"" + csv_filename + "\"")
     with open(csv_filename, mode='w') as csv_file:
-        csv_fieldnames = ['serial_number']
+        csv_fieldnames = ['serial_number', 'closest_match', 'closest_match_percentage']
         csv_writer = csv.writer(csv_file, csv_fieldnames)
         csv_writer.writerow(csv_fieldnames)
         for serial_number in serial_numbers_not_found_in_tablet_usage_data:
-            serial_number_as_list = [serial_number]
-            csv_writer.writerow(serial_number_as_list)
+            # Find the closest match in the tablet usage data
+            closest_match = None
+            closest_match_percentage = None
+            for serial_number_in_tablet_usage_data in serial_numbers_in_tablet_usage_data:
+                serial_match_ratio = get_serial_match_ratio(serial_number, serial_number_in_tablet_usage_data)
+                if (closest_match_percentage is None) or (closest_match_percentage < serial_match_ratio):
+                    closest_match_percentage = serial_match_ratio
+                    closest_match = serial_number_in_tablet_usage_data
+
+            csv_row = [serial_number, closest_match, closest_match_percentage]
+            csv_writer.writerow(csv_row)

team-ONEBILLION/tablet-tracker/serial-numbers-not-found-in-tablet-tracker.csv

@@ -1,42 +1,42 @@
-serial_number
-5A23002711
-5A27000730
-5A27000909
-5A27001802
-5A28000016
-5A28000063
-5A29001362
-5B13001290
-5B20002050
-5B20002381
-5C01000192
-5C01000564
-6109001517
-6111000738
-6111000969
-6111001065
-6111001347
-6111001392
-6111001615
-6111001757
-6111001864
-6115000433
-6115001597
-6115002170
-6116001145
-6116001176
-6116001228
-6116001401
-6116001600
-6116001983
-6116002691
-6116002775
-6118002480
-6118002670
-6118002767
-6118002869
-6118003278
-6118003568
-6126000234
-6129001189
-6130000027
+serial_number,closest_match,closest_match_percentage
+5A23002711,5A23002751,0.8
+5A27000730,5A27000731,0.9
+5A27000909,5A27000924,0.8
+5A27001802,5A27001806,0.9
+5A28000016,5A28000417,0.8
+5A28000063,5A28001065,0.8
+5A29001362,5A29001357,0.8
+5B13001290,5B13001498,0.8
+5B20002050,5B20002255,0.8
+5B20002381,5B20002286,0.8
+5C01000192,5C04000177,0.7
+5C01000564,5C04000177,0.6
+6109001517,6109001591,0.8
+6111000738,6111000756,0.8
+6111000969,6111000974,0.8
+6111001065,6111001060,0.9
+6111001347,6116001343,0.8
+6111001392,6111001096,0.8
+6111001615,6111001619,0.9
+6111001757,6111000756,0.8
+6111001864,6111001869,0.9
+6115000433,6115000461,0.8
+6115001597,6115001496,0.8
+6115002170,6111002172,0.8
+6116001145,6116001147,0.9
+6116001176,6116001178,0.9
+6116001228,6116001229,0.9
+6116001401,6116001402,0.9
+6116001600,6116001606,0.9
+6116001983,6116001981,0.9
+6116002691,6116002694,0.9
+6116002775,6116002578,0.8
+6118002480,6118002407,0.8
+6118002670,6118002672,0.9
+6118002767,6118002764,0.9
+6118002869,6118002850,0.8
+6118003278,6118003275,0.9
+6118003568,6118003567,0.9
+6126000234,6126000273,0.8
+6129001189,6109001185,0.8
+6130000027,6130000023,0.9