-
Notifications
You must be signed in to change notification settings - Fork 0
/
gainFunction.py
131 lines (115 loc) · 8.13 KB
/
gainFunction.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
import numpy as np
import pandas as pd
HUNGRY_MINUTES = 45
MUST = 1
MUST_NOT = 2
def user_inputs_to_gain_function_inputs(diner1_inputs, diner2_inputs, diner3_inputs, rest, meal1_df, meal2_df, meal3_df):
"""
Converts the user inputs to the loss function inputs.
:params diner1_inputs, diner2_inputs, diner3_inputs: user input in format of: [0 - kosher, 1 - vegetarian,
2 - gluten free, 3 - alcohol free, 4 - prefer spicy, 5 - max price, 6 - min rating, 7 - hunger level, 8 - desired cuisines, 9 - day]
:param rest: df with 1 row containing the restaurant
:param meal1_df: df with 1 row containing 1st meal
:param meal2_df: df with 1 row containing 2nd meal
:param meal3_df: df with 1 row containing 3rd meal
:return: a list of inputs for the loss function - [O, M, K, DT, D, RD, R, C, V1, V2, V3, G1, G2, G3, A1, A2, A3, S1, S2, S3, PH1, PH2, PH3, PS1, PS2, PS3]
"""
kosher1, vegetarian1, gluten_free1, alcohol_free1, spicy1, max_price1, rating1, hungry1, cuisines1, weekday = diner1_inputs
kosher2, vegetarian2, gluten_free2, alcohol_free2, spicy2, max_price2, rating2, hungry2, cuisines2, weekday = diner2_inputs
kosher3, vegetarian3, gluten_free3, alcohol_free3, spicy3, max_price3, rating3, hungry3, cuisines3, weekday = diner3_inputs
# Group constraints:
diners_kosher = False if (kosher1 == 0 and kosher2 == 0 and kosher3 == 0) else True
diners_avg_rating = np.mean((rating1, rating2, rating3))
hungry_diners = True if np.sum((hungry1, hungry2, hungry3)) >= 2 else False
rest_cuisines = rest['food categories'].values[0].split('---')
diner1_cui = 1 if len([meal for meal in cuisines1 if meal in rest_cuisines]) > 0 else 0
diner2_cui = 1 if len([meal for meal in cuisines2 if meal in rest_cuisines]) > 0 else 0
diner3_cui = 1 if len([meal for meal in cuisines3 if meal in rest_cuisines]) > 0 else 0
O = 1 if weekday in rest['opening days'].values[0] else 0
M = 1 if ((meal1_df['price'].values[0] + meal2_df['price'].values[0] + meal3_df['price'].values[0]) >= 100) else 0 # TODO replace with order min
K = 0 if diners_kosher and not rest['kosher'].values[0] else 1
DT = rest['delivery estimation [minutes]'].values[0] + rest['prep estimation [minutes]'].values[0]
D = 0 if (hungry_diners and rest['delivery estimation [minutes]'].values[0] + rest['prep estimation [minutes]'].values[0] >= HUNGRY_MINUTES) else 1
RD = rest['rating'].values[0] - diners_avg_rating
R = 1 if diners_avg_rating <= rest['rating'].values[0] else 0
C = diner1_cui + diner2_cui + diner3_cui
# individual constraints:
diner_delivery_cost = + rest['delivery price'].values[0] / 3
price1, price2, price3 = meal1_df['price'].values[0], meal2_df['price'].values[0], meal3_df['price'].values[0]
V1 = 0 if vegetarian1 == MUST and not meal1_df['vegetarian'].values[0] else 1
V2 = 0 if vegetarian2 == MUST and not meal2_df['vegetarian'].values[0] else 1
V3 = 0 if vegetarian3 == MUST and not meal3_df['vegetarian'].values[0] else 1
G1 = 0 if gluten_free1 == MUST and not meal1_df['GF'].values[0] else 1
G2 = 0 if gluten_free2 == MUST and not meal2_df['GF'].values[0] else 1
G3 = 0 if gluten_free3 == MUST and not meal3_df['GF'].values[0] else 1
A1 = 0 if alcohol_free1 == MUST and meal1_df['alcohol_percentage'].values[0] > 0 else 1
A2 = 0 if alcohol_free2 == MUST and meal2_df['alcohol_percentage'].values[0] > 0 else 1
A3 = 0 if alcohol_free3 == MUST and meal3_df['alcohol_percentage'].values[0] > 0 else 1
spicy_meal1, spicy_meal2, spicy_meal3 = meal1_df['spicy'].values[0], meal2_df['spicy'].values[0], meal3_df['spicy'].values[0]
S1 = 0 if (spicy1 == MUST and not spicy_meal1) or (spicy1 == MUST_NOT and spicy_meal1) else 1
S2 = 0 if (spicy2 == MUST and not spicy_meal2) or (spicy2 == MUST_NOT and spicy_meal2) else 1
S3 = 0 if (spicy3 == MUST and not spicy_meal3) or (spicy3 == MUST_NOT and spicy_meal3) else 1
PH1 = 1 if price1 + diner_delivery_cost <= max_price1 else 0
PH2 = 1 if price2 + diner_delivery_cost <= max_price2 else 0
PH3 = 1 if price3 + diner_delivery_cost <= max_price3 else 0
PS1 = max_price1 - price1 if PH1 == 1 else 0
PS2 = max_price2 - price2 if PH2 == 1 else 0
PS3 = max_price3 - price3 if PH3 == 1 else 0
return [O, M, K, DT, D, RD, R, C, V1, V2, V3, G1, G2, G3, A1, A2, A3, S1, S2, S3, PH1, PH2, PH3, PS1, PS2, PS3]
def gain(O, M, K, DT, D, RD, R, C, V1, V2, V3, G1, G2, G3, A1, A2, A3, S1, S2, S3, PH1, PH2, PH3, PS1, PS2, PS3) -> float:
"""
Gain function that can be used to evaluate the fitness of a solution.
based on variables per restaurant (open, minimal order price, kosher, delivery time, rating, cuisines) and per diner (vegetarian, gluten free, alcohol_free, spicy, price).
:param O: (open) - 1 if the restaurant open 0 otherwise
:param M: (minimal order price) - 1 if the meal's combination surpasses the restaurant's minimal order price, 0 otherwise
:param K: (kosher) - 1 if at least one diner eats kosher and the restaurant is kosher or none of the diners eat kosher, 0 otherwise
:param DT: (delivery time) - HUNGRY_MINUTES - (delivery time + preparation time in minutes)
:param D: (delivery) - based on avg hunger level among the group. if hunger level is high - 1 if the meal is ready in less than 30 minutes, 0 otherwise.
if hunger level is low - 1 if the meal is ready in less than 60 minutes, 0 otherwise.
:param RD: (rating difference) - float on a scale of -9 to 9 - the difference between the restaurant's rating and the average rating of the diners
:param R: (rating) - 1 if the restaurant is above avg desired minimal rating among the group or does not have a rating, 0 otherwise
:param C: (cuisines) - int 0-3 according to the amount of diners who prefer a cuisine that the restaurant offers.
:param V1, V2, V3: (vegetarian) 1 if the meal matches the vegetarian desires of the diner, 0 otherwise
:param G1, G2, G3: (gluten free) - 1 if the meal matches the gluten desires of the diner, 0 otherwise
:param A1, A2, A3: (alcohol free) - 1 if the meal matches the alcohol desires of the diner, 0 otherwise
:param S1, S2, S3: (spicy) - 1 if the meal matches the spiciness desires of the diner, 0 otherwise
:param PH1, PH2, PH3: (price hard) - 1 if the meal is lower than the diner's desired maximal meal price, 0 otherwise
:param PS1, PS2, PS3: (price soft) - difference between diner's maximal price and meals price,
0 if the meal's price is higher than the diner's desired maximal meal price
:return: The gain value of the given inputs, according to the desired hard and soft constraints.
"""
"""
hard constraints:
in order to be valid, a combination ob a restaurant and 3 meals must fulfill the following constraints:
- the restaurant must be open (O)
- the sum of prices of all 3 meals must surpass the restaurant's minimal order price (M)
- the meal must match the vegetarian preferences of all 3 diners (V1, V2, V3)
- the meal must match the gluten preferences of all 3 diners (G1, G2, G3)
- the meal must match the alcohol preferences of all 3 diners (A1, A2, A3)
- the meal must be affordable (PH1, PH2, PH3)
"""
hard_constraints = O * M * K * V1 * V2 * V3 * G1 * G2 * G3 * A1 * A2 * A3 * PH1 * PH2 * PH3
if hard_constraints == 0: # if at least 1 hard constraint is not met, return a loss value of 0
return 0
"""
soft constraints:
- delivery time in minutes (DT)
- delivery time matches diners' hunger level (D)
- rating difference (RD)
- rating (R)
- cuisines preferences (C)
- spiciness preferences (S1, S2, S3)
- price differences (PS1, PS2, PS3)
"""
DELIVERY_W = 1
RATING_W = 1
CUISINE_W = 1/3
SPICY_W = 1/2
PRICE_W = 1/2
delivery_gain = (D * DT) / 60 * DELIVERY_W
rating_gain = RD * RATING_W
price_gain = ((PS1 + PS2 + PS3) / 3) / 10 * PRICE_W
cuisine_gain = C * CUISINE_W
spicy_gain = (S1 + S2 + S3) * SPICY_W
sum = (delivery_gain + rating_gain + price_gain + cuisine_gain + spicy_gain)
return sum