I am grouping travellers who travel together, based on the % of trips that they take together. This seems similar to another question ( Group Python list of lists into groups based on overlapping items ) but the conditions are different.
The travellers are only grouped together if they travelled 80% of the trips together . It is ok if the same traveller are in different groups.
Data: (the actual dataset is big and has >1000 trips and travellers)
Traveller Trips
A [Trip_1, Trip_2, Trip_3, Trip_4, Trip_5]
B [Trip_1, Trip_2, Trip_3, Trip_4]
C [Trip_6, Trip_7]
D [Trip_8]
E [Trip_2, Trip_3, Trip_4, Trip_5]
F [Trip_2, Trip_3, Trip_4, Trip_5]
G [Trip_8]
Intended output:
TravelGroup Traveller
Group_1 A
Group_1 B
Group_2 A
Group_2 E
Group_2 F
Group_3 C
Group_4 D
Group_4 G
Note that A and B are in a group; A, E and F are in a group. However, B and C are not in a group because they only have 75% match in the trips taken.
Really appreciate any help here, thank you very much!
df = pd.DataFrame({'Traveller':[*'ABCDE'], 'Trips': [
['Trip_1', 'Trip_2', 'Trip_3', 'Trip_4', 'Trip_5'],
['Trip_1', 'Trip_2', 'Trip_3', 'Trip_4'],
['Trip_1', 'Trip_2'],
['Trip_1'],
['Trip_2', 'Trip_3', 'Trip_4', 'Trip_5']
] })
from itertools import combinations
all_trips = df.explode('Trips')['Trips'].nunique()
all_travelers = set(df.Traveller)
groups, cnt = {'TravelGroup':[], 'Traveller':[]}, 1
for t1, t2 in combinations(df.Traveller, 2):
s1 = df.loc[df.Traveller==t1, 'Trips'].iloc[0]
s2 = df.loc[df.Traveller==t2, 'Trips'].iloc[0]
if len(set(s1).intersection(s2)) / all_trips >= 0.8:
group_name = 'Group_{}'.format(cnt)
groups['TravelGroup'].extend([group_name, group_name])
groups['Traveller'].extend([t1, t2])
cnt += 1
df = pd.DataFrame(groups)
for t in all_travelers.difference(df.Traveller):
group_name = 'Group_{}'.format(cnt)
df.loc[df.shape[0]] = [group_name, t]
cnt += 1
print(df)
Prints:
TravelGroup Traveller
0 Group_1 A
1 Group_1 B
2 Group_2 A
3 Group_2 E
4 Group_3 D
5 Group_4 C
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.