def test_dump():
predictions = [Prediction(None, None, None, None, None)]
algo = AlgoBase()
trainset = Trainset(dict(), *[None] * 8)
with tempfile.NamedTemporaryFile() as tmp_file:
dump(tmp_file.name, predictions, trainset, algo)
def evaluate_model(model: AlgoBase, test_set: [(int, int, float)]) -> dict:
global fit_time
starts = time.time()
predictions = model.test(test_set)
# print("It has been {0} seconds since the evaluation started".format(time.time() - starts))
metrics_dict = {}
metrics_dict['RMSE'] = accuracy.rmse(predictions, verbose=False)
metrics_dict['MAE'] = accuracy.rmse(predictions, verbose=False)
metrics_dict['test_time'] = time.time() - starts
metrics_dict['fit_time'] = fit_time
return metrics_dict
def get_user_recommendation(model: AlgoBase, user_id: int, k: int, data: pandas.DataFrame, movies : pandas.DataFrame
) -> pandas.DataFrame:
"""Makes movie recommendations a user.
Parameters
----------
model : AlgoBase
A trained surprise model
user_id : int
The user for whom the recommendation will be done.
k : int
The number of items to recommend.
data : pandas.DataFrame
The data needed to do the recommendation ( ratings dataframe ).
movies : pandas.DataFrame
The dataframe containing the movies metadata (title, genre, etc)
Returns
-------
pandas.Dataframe
A dataframe with the k movies that will be recommended the user. The dataframe should have the following
columns (movie_name : str, movie_genre : str, predicted_rating : float, true_rating : float)
Notes
-----
- You should create other functions that are used in this one and not put all the code in the same function.
For example to create the final dataframe, instead of implemented all the code
in this function (get_user_recommendation), you can create a new one (create_recommendation_dataframe)
that will be called in this function.
- You can add other arguments to the function if you need to.
"""
data_to_test = create_user_movieId_trueRatings_dataframe(data)
predictions = model.test(data_to_test)
top_n = get_top_n(predictions, n=k)
top_n_df = create_top_n_dataframe(top_n, data_to_test)
recommendation_df = create_recommendation_dataframe(top_n_df, movies, user_id).head(k)
return recommendation_df
#####################################################################################################################
def predict_fn(data: List[InferenceRequest],
model: SurpriseAlgoBase) -> List[surprise.Prediction]:
"""Predict function called for inference"""
return [
model.predict(datum.uid, datum.iid, verbose=True) for datum in data
]
def evaluate_model(model: AlgoBase, test_set: [(int, int, float)]) -> dict:
predictions = model.test(test_set)
metrics_dict = {}
metrics_dict['RMSE'] = accuracy.rmse(predictions, verbose=False)
metrics_dict['MAE'] = accuracy.rmse(predictions, verbose=False)
return metrics_dict
def model_prediction(self, model_class: AlgoBase, test: list) -> list:
pred = model_class.test(test)
return pred
def train(self, param_model: AlgoBase, trainset: Trainset) -> AlgoBase:
param_model.fit(trainset)
return param_model
def make_predictions(model: AlgoBase, test_set: [(int, int, float)]) -> dict:
predictions = model.test(test_set)
return predictions