def rating_of_restaurant(active_user_id,
restaurant_id,
lmu_limit=20,
reco_limit=30,
competitor_limit=20):
# Get details for profile window
user_profile_df = algorithm.user_profile(active_user_id)
# Get first/second order nodes for graph, and info for bar graph comparison
# This is for explore mode, all like nodes are LMU for active users
like_users_df = algorithm.like_users(user_profile_df, active_user_id,
lmu_limit)
# Get third order nodes for graph
recommended_restaurants_df = algorithm.recommended_restaurants(
like_users_df, reco_limit)
#display(recommended_restaurants_df)
# Get all the restaurants name and score inforamtion to build the comparision bar graph
user_graph_dict = algorithm.user_graph(user_id, user_profile_df,
like_users_df,
recommended_restaurants_df)
nodes = user_graph_dict['nodes']
nodes_restaurants = nodes[nodes['order'] == 1][['name', 'score']]
links = user_graph_dict['links'].to_json(orient='records')
nodes = user_graph_dict['nodes'].to_json(orient="records")
graph = {"links": links, "nodes": nodes}
comparison_json = nodes_restaurants.to_json(orient="records")
# Returns details of how user rated restaurant
rating_df = algorithm.user_restaurant_rating(active_user_id, restaurant_id)
rating = rating_df[[
'restaurant', 'user', 'rating', 'date', 'review', 'useful', 'funny',
'cool'
]].astype(str)
rating_json = rating.to_json(orient='records')
#return the likes nodes at the bottom
restaurant_profile_df = algorithm.restaurant_profile(restaurant_id)
#Returns restaurant details with categories and attributes csv lists
like_restaurants_df = algorithm.like_restaurants(restaurant_id,
competitor_limit)
like_restaurants = like_restaurants_df[[
'restaurant', 'restaurant_id', 'compete_score'
]].astype(str).drop_duplicates().head(10)
like_restaurants_json = like_restaurants.to_json(orient="records")
result = {
"graph": graph,
"rating": rating_json,
"comparison": comparison_json,
"likes": like_restaurants_json
}
return result
def likeminded_user_similarity(active_user_id, selected_user_id):
##LMU similarity table
active_user_profile_df = algorithm.user_profile(active_user_id)
active_like_users_df = algorithm.like_users(active_user_profile_df,
active_user_id, lmu_limit)
# Returns information to build first and second order nodes in the graph.
# Also returns name and user_id information to build like user nodes at bottom of screen.
LMU_similarity = active_like_users_df[
active_like_users_df['user_id'] == selected_user_id][[
'shared_val', 'shared_score', 'overlap_val', 'overlap_score',
'rate_sim_val', 'rate_sim_score', 'influence_val',
'influence_score', 'influ_sim_val', 'influ_sim_score', 'lmu_score'
]].astype(str)
LMU_similarity_json = LMU_similarity.to_json(orient='records')
##comparision
# Get third order nodes for graph
recommended_restaurants_df = algorithm.recommended_restaurants(
active_like_users_df, reco_limit)
#display(recommended_restaurants_df)
# Get all the restaurants name and score inforamtion to build the comparision bar graph
user_graph_dict = algorithm.user_graph(active_user_id, user_profile_df,
active_like_users_df,
recommended_restaurants_df)
nodes = user_graph_dict['nodes']
nodes_restaurants = nodes[nodes['order'] == 2][['name', 'score']]
links = user_graph_dict['links'].to_json(orient='records')
nodes = user_graph_dict['nodes'].to_json(orient="records")
graph = {"links": links, "nodes": nodes}
comparison_json = nodes_restaurants.to_json(orient="records")
##likes
selected_user_profile_df = algorithm.user_profile(selected_user_id)
# Get first/second order nodes for graph, and info for like nodes across bottom of page
# This is for explore mode, all like nodes are LMU for active users
like_users_df = algorithm.like_users(selected_user_profile_df,
selected_user_id, lmu_limit)
like_users = like_users_df[['user', 'user_id', 'lmu_score'
]].astype(str).drop_duplicates().head(10)
like_users_json = like_users.to_json(orient="records")
result = {
"graph": graph,
"LMU_similarity": LMU_similarity_json,
"comparison": comparison_json,
"likes": like_users_json
}
return result
def likeminded_user_similarity(active_user_id, selected_user_id):
# Get active user info for other routines
active_user_profile_df = algorithm.user_profile(active_user_id)
#user_profile = active_user_profile_df.astype(str) # Issue #7
#user_profile = active_user_profile_df
# Get first/second order nodes for graph
active_like_users_df = algorithm.like_users(active_user_profile_df,
active_user_id, lmu_limit)
# Pull off details on selected LMU for upper right window
#LMU_similarity = active_like_users_df[active_like_users_df['user_id'] == selected_user_id][
# ['shared_val', 'shared_score', 'overlap_val', 'overlap_score', 'rate_sim_val', 'rate_sim_score',
# 'influence_val', 'influence_score', 'influ_sim_val', 'influ_sim_score', 'lmu_score']].astype(str) # Issue #7
LMU_similarity = active_like_users_df[
active_like_users_df['user_id'] == selected_user_id][[
'shared_val', 'shared_score', 'overlap_val', 'overlap_score',
'rate_sim_val', 'rate_sim_score', 'influence_val',
'influence_score', 'influ_sim_val', 'influ_sim_score', 'lmu_score'
]]
# Get third order nodes for graph
recommended_restaurants_df = algorithm.recommended_restaurants(
active_like_users_df, reco_limit)
# display(recommended_restaurants_df)
# Transform graph info into nodes and links lists
user_graph_dict = algorithm.user_graph(active_user_id,
active_user_profile_df,
active_like_users_df,
recommended_restaurants_df)
# Get info for comparison bar graph
nodes = user_graph_dict['nodes']
nodes_comparison = nodes[nodes['order'] == 2][['name', 'score']]
nodes_comparison.sort_values(by=['score'], ascending=False, inplace=True)
nodes_comparison = nodes_comparison.round({'score': 2})
# Prepare the like user nodes
selected_user_profile_df = algorithm.user_profile(selected_user_id)
like_users_df = algorithm.like_users(selected_user_profile_df,
selected_user_id, lmu_limit)
#like_users_str_df = like_users_df[['user', 'user_id', 'lmu_score']].astype(str).drop_duplicates().head(like_limit) # Issue #7
like_users_str_df = like_users_df[['user', 'user_id', 'lmu_score'
]].drop_duplicates().head(like_limit)
# user_graph_dict, like_users, LMU_similarity, nodes_comparison
return LMU_similarity, nodes_comparison, like_users_str_df
def rating_of_restaurant(active_user_id, restaurant_id):
# Get active user info for other routines
user_profile_df = algorithm.user_profile(active_user_id)
#user_profile = user_profile_df.astype(str) # Issue #7
#user_profile = user_profile_df
# Get first/second order nodes for graph
like_users_df = algorithm.like_users(user_profile_df, active_user_id,
lmu_limit)
# Get third order nodes for graph
recommended_restaurants_df = algorithm.recommended_restaurants(
like_users_df, reco_limit)
# display(recommended_restaurants_df)
# Transform graph info into nodes and links lists
user_graph_dict = algorithm.user_graph(active_user_id, user_profile_df,
like_users_df,
recommended_restaurants_df)
nodes = user_graph_dict['nodes']
nodes_comparison = nodes[nodes['order'] == 1][['name', 'score']]
nodes_comparison.sort_values(by=['score'], ascending=False, inplace=True)
nodes_comparison = nodes_comparison.round({'score': 2})
# Returns details of how active user rated restaurant
rating_df = algorithm.user_restaurant_rating(active_user_id, restaurant_id)
#rating = rating_df[['restaurant', 'user', 'rating', 'date', 'review', 'useful', 'funny', 'cool']].astype(str) # Issue #7
rating = rating_df[[
'restaurant', 'user', 'rating', 'date', 'review', 'useful', 'funny',
'cool'
]]
# return the likes nodes at the bottom # Also use algorithm.like_restaurants for comparison bar graph
like_restaurants_df = algorithm.like_restaurants(restaurant_id,
competitor_limit)
like_restaurants_df['restaurant'] = like_restaurants_df[
'restaurant'].str.encode('utf8')
like_restaurants_str_df = like_restaurants_df[[
'restaurant', 'restaurant_id', 'compete_score'
]].astype(str).drop_duplicates().head(like_limit) # Issue #7
# Return doesn't match View.py which will expect graph_dict, like, rating, comparison
return rating, nodes_comparison, like_restaurants_str_df
def user_profile(user_id):
# Get details for profile window
user_profile_df = algorithm.user_profile(user_id)
user_profile_df['influence'] = np.round(user_profile_df['influence'],
decimals=2)
user_profile_str_df = user_profile_df.astype(str)
user_profile_json = user_profile_str_df.to_json(orient="records")
#display(user_profile_json)
# Get first/second order nodes for graph, and info for like nodes across bottom of page
# This is for explore mode, all like nodes are LMU for active users
like_users_df = algorithm.like_users(user_profile_df, user_id, lmu_limit)
like_users = like_users_df[['user', 'user_id', 'lmu_score'
]].astype(str).drop_duplicates().head(10)
like_users_json = like_users.to_json(orient="records")
#display(like_users_json)
# Get third order nodes for graph
recommended_restaurants_df = algorithm.recommended_restaurants(
like_users_df, reco_limit)
#display(recommended_restaurants_df)
# Transform graph info into nodes and links lists
user_graph_dict = algorithm.user_graph(user_id, user_profile_df,
like_users_df,
recommended_restaurants_df)
links = user_graph_dict['links'].to_json(orient='records')
nodes = user_graph_dict['nodes'].to_json(orient="records")
graph = {"links": links, "nodes": nodes}
#display(graph)
result = {
"profile": user_profile_json,
"graph": graph,
"likes": like_users_json
}
return result
def user_profile(user_id):
# Get details for profile window
user_profile_df = algorithm.user_profile(user_id)
# Note: Full conversion to string with .astype(str) has been commented out throughout API.py to fix issue #7
#user_profile_str_df = user_profile_df.astype(str)
user_profile_str_df = user_profile_df
# Get first/second order nodes for graph, and info for like nodes across bottom of page
like_users_df = algorithm.like_users(user_profile_df, user_id, lmu_limit)
#like_users_str_df = like_users_df[['user', 'user_id', 'lmu_score']].astype(str).drop_duplicates().head(like_limit) # Issue #7
like_users_str_df = like_users_df[['user', 'user_id', 'lmu_score'
]].drop_duplicates().head(like_limit)
# Get third order nodes for graph
recommended_restaurants_df = algorithm.recommended_restaurants(
like_users_df, reco_limit)
# display(recommended_restaurants_df)
# Transform graph info into nodes and links lists
user_graph_dict = algorithm.user_graph(user_id, user_profile_df,
like_users_df,
recommended_restaurants_df)
return user_profile_str_df, user_graph_dict, like_users_str_df
def recommended_restaurant_relevance(active_user_id, selected_restaurant_id):
active_user_profile_df = algorithm.user_profile(active_user_id)
active_like_users_df = algorithm.like_users(active_user_profile_df,
active_user_id, lmu_limit)
recommended_restaurants_df = algorithm.recommended_restaurants(
active_like_users_df, reco_limit)
user_graph_dict = algorithm.user_graph(active_user_id,
active_user_profile_df,
active_like_users_df,
recommended_restaurants_df)
##graph
links = user_graph_dict['links'].to_json(orient='records')
nodes = user_graph_dict['nodes'].to_json(orient="records")
graph = {"links": links, "nodes": nodes}
##relevance
reco_restaurant_relevance = recommended_restaurants_df[
recommended_restaurants_df['restaurant_id'] == selected_restaurant_id]
relevance_df = reco_restaurant_relevance[[
'user', 'lmu_rating', 'lmu_score', 'contribution', 'relevancy'
]]
if (reco_restaurant_relevance.shape[0] == 1):
avg_rating = reco_restaurant_relevance['avg_rating'].values[0]
restaurant_similarity = 5.0
restaurant_contribution = avg_rating * restaurant_similarity
restaurant_relevancy = reco_restaurant_relevance['relevancy'].values[0]
restaurant_df = pd.DataFrame({
'user': ["Restaurant's Average Rating"],
"lmu_rating": [avg_rating],
"lmu_score": [restaurant_similarity],
"contribution": [restaurant_contribution],
"relevancy": [restaurant_relevancy]
})
relevance_df = pd.concat([relevance_df, restaurant_df], axis=0)
relevance = relevance_df.astype(str)
relevance_json = relevance.to_json(orient='records')
##comparison
nodes = user_graph_dict['nodes']
nodes_users = nodes[nodes['order'] == 3][['name', 'score']]
comparison_json = nodes_users.to_json(orient="records")
##likes
selected_restaurant_profile_df = algorithm.restaurant_profile(
selected_restaurant_id)
selected_like_restaurants_df = algorithm.like_restaurants(
selected_restaurant_id, competitor_limit)
like_restaurants = selected_like_restaurants_df[[
'restaurant', 'restaurant_id', 'compete_score'
]].astype(str).drop_duplicates().head(10)
like_restaurants_json = like_restaurants.to_json(orient="records")
result = {
"graph": graph,
"relevance": relevance_json,
"comparison": comparison_json,
"likes": like_restaurants_json
}
return result
def recommended_restaurant_relevance(active_user_id, selected_restaurant_id):
# Get active restaurant info for other routines
active_user_profile_df = algorithm.user_profile(active_user_id)
#user_profile = active_user_profile_df.astype(str) # Issue #7
#user_profile = active_user_profile_df
# Get first/second order nodes for graph
active_like_users_df = algorithm.like_users(active_user_profile_df,
active_user_id, lmu_limit)
# Get third order nodes for graph
recommended_restaurants_df = algorithm.recommended_restaurants(
active_like_users_df, reco_limit)
# Transform graph info into nodes and links lists
user_graph_dict = algorithm.user_graph(active_user_id,
active_user_profile_df,
active_like_users_df,
recommended_restaurants_df)
# Pull off details on selected recommended restaurant relevance for upper right window
reco_restaurant_relevance = recommended_restaurants_df[
recommended_restaurants_df['restaurant_id'] == selected_restaurant_id]
relevance_df = reco_restaurant_relevance[[
'user', 'lmu_rating', 'lmu_score', 'contribution', 'relevancy'
]]
if (reco_restaurant_relevance.shape[0] == 1):
avg_rating = reco_restaurant_relevance['avg_rating'].values[0]
restaurant_similarity = 5.0
restaurant_contribution = avg_rating * restaurant_similarity
restaurant_relevancy = reco_restaurant_relevance['relevancy'].values[0]
restaurant_df = pd.DataFrame({
'user': ["Restaurant's Average Rating"],
"lmu_rating": [avg_rating],
"lmu_score": [restaurant_similarity],
"contribution": [restaurant_contribution],
"relevancy": [restaurant_relevancy]
})
relevance_df = pd.concat([relevance_df, restaurant_df], axis=0)
#relevance = relevance_df.astype(str) # Issue #7
relevance = relevance_df
# Get info for comparison bar graph
nodes = user_graph_dict['nodes']
nodes_comparison = nodes[nodes['order'] == 3][['name', 'score'
]] # nodes_comparison
nodes_comparison.sort_values(by=['score'], ascending=False, inplace=True)
nodes_comparison = nodes_comparison.round({'score': 2})
# Prepare the like restaurant nodes
selected_like_restaurants_df = algorithm.like_restaurants(
selected_restaurant_id,
competitor_limit)[['restaurant', 'restaurant_id', 'compete_score']]
selected_like_restaurants_df['restaurant'] = selected_like_restaurants_df[
'restaurant'].str.encode('utf8')
selected_like_restaurants_df[
'restaurant_id'] = selected_like_restaurants_df[
'restaurant_id'].str.encode('utf8')
selected_like_restaurants_df[
'compete_score'] = selected_like_restaurants_df[
'compete_score'].astype(str)
selected_like_restaurants_df = selected_like_restaurants_df.drop_duplicates(
).head(like_limit) # Issue #7
return relevance, nodes_comparison, selected_like_restaurants_df