def read_item(userId: str, itemID: str):
userID = userId
itemID = itemID
df1 = pd.DataFrame(data={"userID": [int(userID)], "itemID": [str(itemID)]})
bpr = pickle.load(open("lightGCN_model.p", "rb"))
#bpr = pickle.load( open( "trained_model.p", "rb" ))
#bpr = pickle.load( open( "new_model.p", "rb" ))
with Timer() as t:
foo = predict_ranking(bpr,
df1,
usercol='userID',
itemcol='itemID',
remove_seen=True)
print("Took {} seconds for prediction.".format(t))
foo["userID"] = foo["userID"].apply(lambda x: round(x))
foo["itemID"] = foo["itemID"].apply(lambda x: (x))
result = foo[foo["userID"] == int(userID)].sort_values(
by='prediction', ascending=False).head(10)
result = result.drop(columns=["prediction", "userID"], axis=1)
#data = {
# "data":result
#}
#return json.dumps(data)
return result.to_json()
def recommend_k_bpr(model, test, train):
with Timer() as t:
topk_scores = predict_ranking(
model,
train,
usercol=DEFAULT_USER_COL,
itemcol=DEFAULT_ITEM_COL,
predcol=DEFAULT_PREDICTION_COL,
remove_seen=True,
)
return topk_scores, t
def test_predict_ranking(rating_true):
train_set = cornac.data.Dataset.from_uir(rating_true.itertuples(index=False), seed=42)
bpr = cornac.models.BPR(k=100, max_iter=10000, seed=42).fit(train_set)
preds = predict_ranking(bpr, rating_true, remove_seen=False)
n_users = len(rating_true["userID"].unique())
n_items = len(rating_true["itemID"].unique())
assert preds.shape[0] == n_users * n_items
assert set(preds.columns) == {"userID", "itemID", "prediction"}
assert preds["userID"].dtypes == rating_true["userID"].dtypes
assert preds["itemID"].dtypes == rating_true["itemID"].dtypes
# perfect ranking achieved
assert 1e-10 > 1 - ndcg_at_k(rating_true, preds)
assert 1e-10 > 1 - recall_at_k(rating_true, preds)
def read_item(userId: str, itemID: str):
userID = userId
itemID = itemID
df1 = pd.DataFrame(data={"userID": [int(userID)], "itemID": [int(itemID)]})
svd = pickle.load(open("SVD_pickle.p", "rb"))
with Timer() as t:
foo = predict_ranking(svd,
df1,
usercol='userID',
itemcol='itemID',
remove_seen=True)
print("Took {} seconds for prediction.".format(t))
foo["userID"] = foo["userID"].apply(lambda x: round(x))
foo["itemID"] = foo["itemID"].apply(lambda x: round(x))
result = foo[foo["userID"] == int(userID)].sort_values(
by='prediction', ascending=False).head(10)
result = result.drop(columns=["prediction", "userID"], axis=1)
return result.to_json()