backend/passfinder/recomendations/service/service.py

573 lines
18 KiB
Python
Raw Normal View History

2023-05-21 13:37:21 +03:00
from annoy import AnnoyIndex
from .mapping.mapping import *
from .models.models import *
2023-05-25 17:12:39 +03:00
from passfinder.events.models import Event, Region, Hotel, BasePoint, City, Restaurant
from passfinder.events.api.serializers import HotelSerializer, EventSerializer, ResaurantSerializer, ObjectRouteSerializer
from passfinder.recomendations.models import *
2023-05-24 17:58:51 +03:00
from random import choice, sample
2023-05-21 17:13:16 +03:00
from collections import Counter
2023-05-22 22:37:54 +03:00
from passfinder.users.models import User
from collections.abc import Iterable
2023-05-23 17:51:01 +03:00
from django.db.models import Q
from geopy.distance import geodesic as GD
from datetime import timedelta, time, datetime
2023-05-25 17:12:39 +03:00
from gevent.pool import Pool
2023-05-21 13:37:21 +03:00
2023-05-23 17:51:01 +03:00
def get_nearest_(instance_model, model_type, mapping, rev_mapping, nearest_n, ml_model):
2023-05-21 13:37:21 +03:00
how_many = len(Event.objects.filter(type=model_type))
2023-05-23 17:51:01 +03:00
index = rev_mapping[instance_model.oid]
2023-05-21 13:37:21 +03:00
nearest = ml_model.get_nns_by_item(index, len(mapping))
res = []
for i in range(how_many):
try:
res.append(Event.objects.get(oid=mapping[nearest[i]]))
except Event.DoesNotExist:
...
if len(res) == nearest_n:
break
2023-05-21 13:37:21 +03:00
return res
def nearest_attraction(attraction, nearest_n):
return get_nearest_(
attraction,
"attraction",
attraction_mapping,
rev_attraction_mapping,
nearest_n,
attracion_model,
)
2023-05-21 13:37:21 +03:00
2023-05-24 17:58:51 +03:00
def nearest_mus(museum, nearest_n):
return get_nearest_(
museum,
"museum",
mus_mapping,
rev_mus_mapping,
nearest_n,
mus_model
)
2023-05-21 13:37:21 +03:00
def nearest_movie(movie, nearest_n):
return get_nearest_(
movie, "movie", cinema_mapping, rev_cinema_mapping, nearest_n, cinema_model
)
2023-05-21 13:37:21 +03:00
def nearest_plays(play, nearest_n):
return get_nearest_(
play, "plays", plays_mapping, rev_plays_mapping, nearest_n, plays_model
)
2023-05-21 13:37:21 +03:00
def nearest_excursion(excursion, nearest_n):
return get_nearest_(
excursion,
"excursion",
excursion_mapping,
rev_excursion_mapping,
nearest_n,
excursion_model,
)
2023-05-21 13:37:21 +03:00
def nearest_concert(concert, nearest_n):
return get_nearest_(
concert,
"concert",
concert_mapping,
rev_concert_mapping,
nearest_n,
concert_model,
)
2023-05-21 17:13:16 +03:00
def get_nearest_event(event, nearest_n):
if event.type == "plays":
2023-05-21 17:13:16 +03:00
return nearest_plays(event, nearest_n)
if event.type == "concert":
2023-05-21 17:13:16 +03:00
return nearest_concert(event, nearest_n)
if event.type == "movie":
2023-05-21 17:13:16 +03:00
return nearest_movie(event, nearest_n)
2023-05-24 17:58:51 +03:00
if event.type == 'museum':
return nearest_mus(event, nearest_n)
if event.type == 'attraction':
return nearest_attraction(event, nearest_n)
2023-05-21 17:13:16 +03:00
def update_preferences_state(user, event, direction):
pref = UserPreferences.objects.get(user=user)
if direction == "left":
if event.type == "plays":
2023-05-21 17:13:16 +03:00
pref.unpreffered_plays.add(event)
if event.type == "movie":
2023-05-21 17:13:16 +03:00
pref.unpreffered_movies.add(event)
if event.type == "concert":
2023-05-21 17:13:16 +03:00
pref.unpreferred_concerts.add(event)
else:
if event.type == "plays":
2023-05-21 17:13:16 +03:00
pref.preffered_plays.add(event)
if event.type == "movie":
2023-05-21 17:13:16 +03:00
pref.preffered_movies.add(event)
if event.type == "concert":
2023-05-21 17:13:16 +03:00
pref.preferred_concerts.add(event)
pref.save()
def get_next_tinder(user, prev_event, prev_direction):
pref = UserPreferences.objects.get(user=user)
print(prev_event.type, len(pref.preferred_concerts.all()))
if prev_direction == "left":
if prev_event.type == "plays" and len(pref.unpreffered_plays.all()) <= 2:
2023-05-21 17:13:16 +03:00
candidates = nearest_plays(prev_event, 100)
# print(candidates, type(candidates), len(Event.objects.filter(type='plays')))
return candidates[-1]
if prev_event.type == "movie" and len(pref.unpreffered_movies.all()) <= 2:
2023-05-21 17:13:16 +03:00
candidates = nearest_movie(prev_event, 100)
return candidates[-1]
if prev_event.type == "concert" and len(pref.unpreferred_concerts.all()) <= 2:
2023-05-21 17:13:16 +03:00
candidates = nearest_concert(prev_event, 100)
return candidates[-1]
if prev_direction == "right":
if prev_event.type == "plays" and len(pref.preffered_plays.all()) < 2:
2023-05-21 17:13:16 +03:00
candidates = nearest_plays(prev_event, 2)
return candidates[1]
if prev_event.type == "movie" and len(pref.preffered_movies.all()) < 2:
2023-05-21 17:13:16 +03:00
candidates = nearest_movie(prev_event, 2)
return candidates[1]
if prev_event.type == "concert" and len(pref.preferred_concerts.all()) < 2:
2023-05-21 17:13:16 +03:00
candidates = nearest_concert(prev_event, 2)
return candidates[1]
if prev_event.type == "plays":
if not len(pref.preffered_movies.all()) and not len(
pref.unpreffered_movies.all()
):
return choice(Event.objects.filter(type="movie"))
if not len(pref.preferred_concerts.all()) and not len(
pref.unpreferred_concerts.all()
):
return choice(Event.objects.filter(type="concert"))
if prev_event.type == "movie":
if not len(pref.preffered_plays.all()) and not len(
pref.unpreffered_plays.all()
):
return choice(Event.objects.filter(type="plays"))
if not len(pref.preferred_concerts.all()) and not len(
pref.unpreferred_concerts.all()
):
return choice(Event.objects.filter(type="concert"))
if prev_event.type == "concert":
if not len(pref.preffered_plays.all()) and not len(
pref.unpreffered_plays.all()
):
return choice(Event.objects.filter(type="plays"))
if not len(pref.preffered_movies.all()) and not len(
pref.unpreffered_movies.all()
):
return choice(Event.objects.filter(type="movie"))
2023-05-21 17:13:16 +03:00
return None
2023-05-21 17:13:16 +03:00
def rank_candidates(candidates_list, negative_candidates_list):
flatten_c_list = []
ranks = {}
flatten_negatives = []
for negative in negative_candidates_list:
flatten_negatives.extend(negative)
2023-05-21 17:13:16 +03:00
for lst in candidates_list:
flatten_c_list.extend(lst)
for cand in lst:
ranks.update({cand: {"rank": 0, "lst": lst}})
2023-05-21 17:13:16 +03:00
cnt = Counter(flatten_c_list)
for candidate, how_many in cnt.most_common(len(flatten_c_list)):
ranks[candidate]["rank"] = how_many * (
len(ranks[candidate]["lst"]) - ranks[candidate]["lst"].index(candidate)
)
2023-05-21 17:13:16 +03:00
res = []
for cand in ranks.keys():
res.append((ranks[cand]["rank"], cand))
return list(
filter(
lambda x: x[1] not in flatten_negatives, sorted(res, key=lambda x: -x[0])
)
)
2023-05-21 17:13:16 +03:00
def get_personal_recommendation(prefer, unprefer):
candidates = []
negative_candidates = []
for rec in prefer:
candidates.append(list(map(lambda x: x.oid, get_nearest_event(rec, 10)[1:])))
2023-05-21 17:13:16 +03:00
for neg in unprefer:
negative_candidates.append(
list(map(lambda x: x.oid, get_nearest_event(neg, 10)[1:]))
)
2023-05-21 17:13:16 +03:00
ranked = rank_candidates(candidates, negative_candidates)
return list(map(lambda x: (x[0], Event.objects.get(oid=x[1])), ranked[0:5]))
def get_personal_plays_recommendation(user):
pref = UserPreferences.objects.get(user=user)
prefer = pref.preffered_plays.all()
unprefer = pref.unpreffered_plays.all()
return get_personal_recommendation(prefer, unprefer)
def get_personal_concerts_recommendation(user):
pref = UserPreferences.objects.get(user=user)
prefer = pref.preferred_concerts.all()
unprefer = pref.unpreferred_concerts.all()
return get_personal_recommendation(prefer, unprefer)
def get_personal_movies_recommendation(user):
pref = UserPreferences.objects.get(user=user)
prefer = pref.preffered_movies.all()
unprefer = pref.unpreffered_movies.all()
return get_personal_recommendation(prefer, unprefer)
2023-05-22 22:37:54 +03:00
def dist_func(event1: Event, event2: Event):
2023-05-25 17:12:39 +03:00
cords1 = [event1.lat, event1.lon]
cords2 = [event2.lat, event2.lon]
try:
dist = GD(cords1, cords2).km
return dist
except:
return 1000000
#return (event1.lon - event2.lon) ** 2 + (event1.lat - event2.lat) ** 2
2023-05-22 22:37:54 +03:00
def generate_nearest():
NearestEvent.objects.all().delete()
all_events = list(Event.objects.all())
for i, event in enumerate(Event.objects.all()):
event_all_events = list(
sorted(all_events.copy(), key=lambda x: dist_func(event, x))
)
2023-05-22 22:37:54 +03:00
nearest = NearestEvent.objects.create(event=event)
nearest.nearest.set(event_all_events[0:100])
nearest.save()
if i % 100 == 0:
print(i)
2023-05-23 17:51:01 +03:00
def generate_hotel_nearest():
NearestHotel.objects.all().delete()
all_events = list(Event.objects.all())
hotels = list(Hotel.objects.all())
for i, hotel in enumerate(hotels):
event_all_events = list(
sorted(all_events.copy(), key=lambda x: dist_func(hotel, x))
)
2023-05-23 17:51:01 +03:00
nearest = NearestHotel.objects.create(hotel=hotel)
nearest.nearest_events.set(event_all_events[0:100])
if i % 100 == 0:
print(i)
2023-05-25 17:12:39 +03:00
def generate_nearest_restaurants():
rests = list(Restaurant.objects.all())
for i, event in enumerate(Event.objects.all()):
sorted_rests = list(sorted(rests.copy(), key=lambda x: dist_func(x, event)))
nr = NearestRestaurantToEvent.objects.create(event=event)
nr.restaurants.set(sorted_rests[0:20])
nr.save()
if i % 100 == 0:
print(i)
for i, hotel in enumerate(Hotel.objects.all()):
sorted_rests = list(sorted(rests.copy(), key=lambda x: dist_func(x, hotel)))
nr = NearestRestaurantToHotel.objects.create(hotel=hotel)
nr.restaurants.set(sorted_rests[0:20])
nr.save()
if i % 100 == 0:
print(i)
2023-05-23 18:27:14 +03:00
def match_points():
regions = list(City.objects.all())
for i, point in enumerate(Event.objects.all()):
s_regions = list(sorted(regions.copy(), key=lambda x: dist_func(point, x)))
point.city = s_regions[0]
point.save()
if i % 10 == 0:
print(i)
2023-05-23 17:51:01 +03:00
def calculate_mean_metric(
favorite_events: Iterable[Event],
target_event: Event,
model: AnnoyIndex,
rev_mapping,
):
2023-05-22 22:37:54 +03:00
if not len(favorite_events):
return 100000
2023-05-22 22:37:54 +03:00
dists = []
2023-05-23 17:51:01 +03:00
target_event_idx = rev_mapping[target_event.oid]
2023-05-22 22:37:54 +03:00
for fav in favorite_events:
2023-05-23 17:51:01 +03:00
dists.append(model.get_distance(rev_mapping[fav.oid], target_event_idx))
2023-05-22 22:37:54 +03:00
return sum(dists) / len(dists)
def calculate_favorite_metric(event: Event, user: User):
pref = UserPreferences.objects.get(user=user)
if event.type == "plays":
2023-05-22 22:37:54 +03:00
preferred = pref.preffered_plays.all()
return calculate_mean_metric(preferred, event, plays_model, rev_plays_mapping)
if event.type == "concert":
2023-05-22 22:37:54 +03:00
preferred = pref.preferred_concerts.all()
return calculate_mean_metric(
preferred, event, concert_model, rev_concert_mapping
2023-05-22 22:37:54 +03:00
)
if event.type == "movie":
2023-05-22 22:37:54 +03:00
preferred = pref.preffered_movies.all()
return calculate_mean_metric(preferred, event, cinema_model, rev_cinema_mapping)
2023-05-24 17:58:51 +03:00
if event.type == 'attraction':
preferred = pref.prefferred_attractions.all()
return calculate_mean_metric(preferred, event, attracion_model, rev_attraction_mapping)
if event.type == 'museum':
preferred = pref.prefferred_museums.all()
return calculate_mean_metric(preferred, event, mus_model, rev_mus_mapping)
2023-05-22 22:37:54 +03:00
return 1000000
def get_nearest_favorite(
events: Iterable[Event], user: User, exclude_events: Iterable[Event] = []
):
2023-05-25 17:12:39 +03:00
print(events)
2023-05-23 17:51:01 +03:00
first_event = None
for candidate in events:
if candidate not in exclude_events:
2023-05-23 17:51:01 +03:00
first_event = candidate
break
2023-05-25 17:12:39 +03:00
if first_event is None:
result = events[0]
else:
result = first_event
result_min = 1000000
2023-05-22 22:37:54 +03:00
for event in events:
if event in exclude_events:
continue
2023-05-22 22:37:54 +03:00
local_min_metric = calculate_favorite_metric(event, user)
if local_min_metric < result_min:
result_min = local_min_metric
result = event
return result
2023-05-23 17:51:01 +03:00
def filter_hotel(region: Region, user: User, stars: Iterable[int]):
2023-05-23 23:38:45 +03:00
hotels = Hotel.objects.filter(city=region)
2023-05-23 17:51:01 +03:00
return choice(hotels)
def time_func(km_distance: float):
return timedelta(minutes=(km_distance) / (4.0 / 60))
def generate_route(point1: BasePoint, point2: BasePoint):
distance = dist_func(point1, point2)
time = time_func(distance)
return {
"type": "transition",
"distance": distance,
2023-05-25 17:12:39 +03:00
"time": time.seconds,
2023-05-23 17:51:01 +03:00
}
def generate_point(point: BasePoint):
2023-05-25 17:12:39 +03:00
event_data = ObjectRouteSerializer(point).data
2023-05-23 17:51:01 +03:00
return {
"type": "point",
2023-05-25 17:12:39 +03:00
"point": event_data,
"point_type": "point",
"time": timedelta(minutes=90+choice(range(-10, 90, 10))).seconds
2023-05-23 17:51:01 +03:00
}
2023-05-25 17:12:39 +03:00
def generate_restaurant(point: BasePoint):
rest_data = ObjectRouteSerializer(point).data
return {
"type": "point",
"point": rest_data,
"point_type": "restaurant",
"time": timedelta(minutes=90+choice(range(-10, 90, 10))).seconds
}
2023-05-22 22:37:54 +03:00
2023-05-23 17:51:01 +03:00
2023-05-25 17:12:39 +03:00
def generate_multiple_tours(user: User, city: City, start_date: datetime.date, end_date: datetime.date):
hotels = sample(list(Hotel.objects.filter(city=city)), 5)
pool = Pool(5)
return pool.map(generate_tour, [(user, start_date, end_date, hotel) for hotel in hotels])
def generate_tour(user: User, city: City, start_date: datetime.date, end_date: datetime.date):
print("start hotel")
hotel = choice(list(Hotel.objects.filter(city=city)))
print("end hotel")
current_date = start_date
paths, points = [], []
while current_date < end_date:
print("start day gen")
local_points, local_paths = generate_path(user, points, hotel)
points.extend(local_points)
paths.append(
{
'date': current_date,
'paths': local_paths
}
)
print("end day gen")
2023-05-22 22:37:54 +03:00
2023-05-25 17:12:39 +03:00
current_date += timedelta(days=1)
return paths, points
2023-05-22 22:37:54 +03:00
2023-05-25 17:12:39 +03:00
def generate_hotel(hotel: Hotel):
hotel_data = ObjectRouteSerializer(hotel).data
return {
"type": "point",
"point": hotel_data,
"point_type": "hotel",
"time": timedelta(minutes=90+choice(range(-10, 90, 10))).seconds
}
def generate_path(user: User, disallowed_points: Iterable[BasePoint], hotel: Hotel):
# region_events = Event.objects.filter(region=region)
#candidates = NearestHotel.objects.get(hotel=hotel).nearest_events.all()
allowed_types = ['museum', 'attraction']
print("start start point gen")
start_point = NearestRestaurantToHotel.objects.get(hotel=hotel).restaurants.first()
print("end start point gen")
print("start first cand gen")
candidates = list(filter(lambda x: x.type in allowed_types, map(lambda x: x.event, start_point.nearestrestauranttoevent_set.all()[0:100])))
print("end first cand gen")
points = [start_point]
path = [
generate_hotel(hotel),
generate_route(start_point, hotel),
generate_restaurant(points[-1])
]
2023-05-23 17:51:01 +03:00
start_time = datetime.combine(datetime.now(), time(hour=10))
2023-05-25 17:12:39 +03:00
how_many_eat = 1
2023-05-23 17:51:01 +03:00
while start_time.hour < 22:
2023-05-25 17:12:39 +03:00
print(start_time)
if (start_time.hour > 13 and how_many_eat == 1) or (start_time.hour > 20 and how_many_eat == 2):
print("start rest event gen")
point = NearestRestaurantToEvent.objects.get(event=points[-1]).restaurants.all()[0]
points.append(point)
candidates = list(filter(lambda x: x.type in allowed_types, map(lambda x: x.event, point.nearestrestauranttoevent_set.all()[0:100])))
if not len(candidates):
candidates = list(map(lambda x: x.event, point.nearestrestauranttoevent_set.all()[0:100]))
path.append(generate_restaurant(points[-1]))
start_time += timedelta(seconds=path[-1]['time'])
how_many_eat += 1
print("start rest event gen")
continue
2023-05-25 17:12:39 +03:00
if start_time.hour > 17:
allowed_types = ['play', 'concert', 'movie']
print("start events gen")
if candidates is None:
candidates = NearestEvent.objects.get(event=points[-1]).nearest.filter(type__in=allowed_types)
if not len(candidates):
candidates = NearestEvent.objects.get(event=points[-1]).nearest.all()
print("end events gen")
print("start events select")
try:
print(points)
points.append(get_nearest_favorite(candidates, user, points + disallowed_points))
except AttributeError:
points.append(get_nearest_favorite(candidates, user, points))
print("end events select")
print("start route gen")
2023-05-23 17:51:01 +03:00
transition_route = generate_route(points[-1], points[-2])
2023-05-25 17:12:39 +03:00
start_time += timedelta(seconds=transition_route["time"])
2023-05-23 17:51:01 +03:00
point_route = generate_point(points[-1])
2023-05-25 17:12:39 +03:00
start_time += timedelta(seconds=point_route["time"])
2023-05-23 17:51:01 +03:00
path.extend([transition_route, point_route])
2023-05-25 17:12:39 +03:00
candidates = None
print("end route gen")
return points, path
2023-05-24 17:58:51 +03:00
def calculate_distance(sample1: Event, samples: Iterable[Event], model: AnnoyIndex, rev_mapping):
metrics = []
for sample in samples:
metrics.append(model.get_distance(rev_mapping[sample1.oid], rev_mapping[sample.oid]))
return sum(metrics) / len(metrics)
def get_onboarding_attractions():
sample_attractions = sample(list(Event.objects.filter(type='attraction')), 200)
first_attraction = choice(sample_attractions)
attractions = [first_attraction]
while len(attractions) < 10:
mx_dist = 0
mx_attraction = None
for att in sample_attractions:
if att in attractions: continue
local_dist = calculate_distance(
att,
attractions,
attracion_model,
rev_attraction_mapping
)
if local_dist > mx_dist:
mx_dist = local_dist
mx_attraction = att
attractions.append(mx_attraction)
return attractions
def get_onboarding_hotels(stars=Iterable[int]):
return sample(list(Hotel.objects.filter(stars__in=stars)), 10)