def get_ranked_ingreds_from_cooc(ingred):
ingreds = get_json('all_ingreds_filtered.json')
ingred_to_ix = {k: v for v, k in enumerate(ingreds)}
ix_to_ingred = {v: k for v, k in enumerate(ingreds)}
cooc = np.array(get_json('cooc.json'))
ingred_ix = ingred_to_ix[ingred]
ranked_ixs = np.argsort(cooc[ingred_ix])
ranked_ixs = np.flip(ranked_ixs)
ranked_ingreds = {}
for ranked_ix in ranked_ixs:
cooccurrences = cooc[ingred_ix, ranked_ix]
if cooccurrences == 0:
break
ranked_ingreds[ix_to_ingred[ranked_ix]] = cooccurrences
return ranked_ingreds
def make_recipe_matrix():
'''2D matrix whose rows are ingredients and cols are recipes titles.
A 1 denotes the occurence of an ingredient in a given recipe.'''
ingreds = get_json('all_ingreds_filtered.json')
recipes = get_json('recipe_data_filtered.json')
titles = []
for recipe in recipes:
titles.append(recipe['title'])
df = pd.DataFrame(0, ingreds, titles)
ingreds = set(ingreds)
for recipe in recipes:
recipe_ingreds = set(recipe['ingreds'])
matches = recipe_ingreds & ingreds
if len(matches) > 0:
df.loc[list(matches), recipe['title']] += 1
return df.to_numpy()
def write_recipe_matrix(outfile='recipe_matrix.json'):
'''2D matrix whose rows are ingredients and cols are recipes.
A 1 denotes the occurence of an ingredient in a given recipe.'''
ingreds = helper.get_json('all_ingreds_filtered.json')
recipes = helper.get_json('recipe_data_filtered.json')
titles = []
for recipe in recipes:
titles.append(recipe['title'])
df = pd.DataFrame(0, ingreds, titles)
ingreds = set(ingreds)
for recipe in recipes:
recipe_ingreds = set(recipe['ingreds'])
matches = recipe_ingreds & ingreds
if len(matches) > 0:
df.loc[list(matches), recipe['title']] = 1
data = df.to_numpy()
data = data.tolist()
helper.write_json(data, outfile, 'w')
def write_all_ingreds(recipe_file_name, ingred_file_name):
"""Save json of all ingreds in recipe_file_name to ingred_file_name."""
data = helper.get_json(recipe_file_name)
ingreds = []
for recipe in data:
ingreds.append(recipe['ingreds'])
ingreds = [ingred for sublist in ingreds for ingred in sublist]
ingreds = list(set(ingreds))
ingreds.sort()
helper.write_json(ingreds, ingred_file_name, 'w')
return ingreds
def write_recipe_data_filtered(infile, outfile):
"""Filter recipes from infile and save to outfile as json."""
data = helper.get_json(infile)
with open('approved_ingreds', 'r', encoding="utf8") as f:
approved_ingreds = set(f.read().splitlines())
ingred_filters = generate_ingred_filters(approved_ingreds)
# Remove duplicate recipes
df = pd.DataFrame(data)
df_unique = df[~df['title'].duplicated()]
data = df_unique.to_dict('records')
for recipe in data:
filtered_ingreds = filter_naive(recipe['ingreds'], ingred_filters)
recipe['ingreds'] = filtered_ingreds
helper.write_json(data, outfile, 'w')
def write_all_ingreds_lemma(infile='all_ingreds_filtered.json',
outfile='static/all_ingreds_lemma.json'):
"""Save json of lemmatization of ingreds in infile to outfile."""
ingreds = helper.get_json(infile)
ingreds = [lemmatize(ingred) for ingred in ingreds]
helper.write_json(ingreds, outfile, 'w')
c_property = df['c_property']
rew = df['rew']
rew_clean = df['rew_clean']
rew_latlong = df['latlong']
# Load the model
model = get_model()
gbr = model['gbr']
f1 = model['f1n']
f2 = model['f2n']
f3 = model['f3n']
f4 = model['f4n']
stackedmodel = model['stackedmodel']
geo_json = get_json()
recommender = get_recommender()
#variables
neighbourhoods = list(np.array(area['Neighbourhood_Name']))
n_values = list(np.array(area['N_Value']))
options = [{
'label': name,
'value': value
} for name, value in zip(neighbourhoods, n_values)]
mapbox_access_token = "pk.eyJ1IjoiamFja3AiLCJhIjoidGpzN0lXVSJ9.7YK6eRwUNFwd3ODZff6JvA"
year = 2020
GoogleMaps(app, key="AIzaSyAvzm0x2kmJoeqX6VY83lMa8pPRDDjH4sY")
cols = list(rew.columns)
type_list = cols[cols.index('House'):cols.index('Multifamily') + 1]
import http
import random
from flask import Flask, jsonify, render_template, request, session
import matrix
from helper import get_json
app = Flask(__name__)
app.config.from_object('config')
VERSION_STR = '?v=0.61'
ALL_INGREDS = [
ingred.lower() for ingred in get_json('./static/all_ingreds_lemma.json')
]
# TODO:
# Try render_template instead of js stuff? Just use js for fetch?
# Or even move that straight to html with method="POST" action="/search"?
# - Get seperate JQuery source seperate from datatables bundle.
# load JQuery first, then typeahead, then datatables
# for faster load time.
# - allow multiple inputs
# Emojis, anyone?
@app.route("/")
def index():
init_session_vars()
remove_invalid_session_ingreds()
cur_ingreds = get_session_var('cur_ingreds')
import matplotlib.pyplot as plt
import networkx as nx
import numpy as np
from helper import get_json
if __name__ == "__main__":
cooc = get_json('cooc.json')
ingreds = get_json('all_ingreds_filtered.json')
G = nx.from_numpy_array(np.array(cooc))
nx.draw(G)
plt.show()
def main():
ingreds = 'onion'
recipe_matrix = np.array(get_json('recipe_matrix.json'))
all_ingreds = get_json('all_ingreds_filtered.json')
get_ranked_ingreds_naive(ingreds, recipe_matrix, all_ingreds)
import numpy as np
from sklearn.metrics.pairwise import cosine_similarity as cs
from helper import get_json, timer
'''2D matrix whose rows are ingredients and cols are recipes.
A 1 denotes the occurence of an ingredient in a given recipe.'''
RECIPE_MATRIX = np.array(get_json('recipe_matrix.json'))
INGRED_AXIS = 0
RECIPE_AXIS = 1
ALL_INGREDS = get_json('all_ingreds_filtered.json')
INGRED_TO_IX = {k: i for i, k in enumerate(ALL_INGREDS)}
IX_TO_INGRED = {i: k for i, k in enumerate(ALL_INGREDS)}
RECIPE_DATA = get_json('recipe_data_filtered.json')
IX_TO_RECIPE = {i: (r['title'], r['url']) for i, r in enumerate(RECIPE_DATA)}
def get_recommended(input_ingreds):
if isinstance(input_ingreds, str):
input_ingreds = [input_ingreds]
match_recipe_ixs = get_match_recipe_ixs(input_ingreds)
ranked_ingreds = get_ranked_ingreds(input_ingreds, match_recipe_ixs)
match_recipes = get_match_recipes(match_recipe_ixs)
return ranked_ingreds, match_recipes
def get_ranked_ingreds(input_ingreds, match_recipe_ixs):