src.fairreckitlib.model.pipeline.prediction_pipeline

This module contains a model pipeline that predicts known item ratings.

Classes:

PredictionPipeline: can batch predictions from multiple models for a specific API.

This program has been developed by students from the bachelor Computer Science at Utrecht University within the Software Project course. © Copyright Utrecht University (Department of Information and Computing Sciences)

View Source

 1"""This module contains a model pipeline that predicts known item ratings.
 2
 3Classes:
 4
 5    PredictionPipeline: can batch predictions from multiple models for a specific API.
 6
 7This program has been developed by students from the bachelor Computer Science at
 8Utrecht University within the Software Project course.
 9© Copyright Utrecht University (Department of Information and Computing Sciences)
10"""
11
12from typing import List
13
14import pandas as pd
15
16from ..algorithms.base_predictor import BasePredictor
17from .model_pipeline import ModelPipeline
18
19
20class PredictionPipeline(ModelPipeline):
21    """Prediction Pipeline that computes user/item rating predictions.
22
23    The (user,item) prediction will be computed and for each pair that is present in the test set.
24    """
25
26    def load_test_set_users(self) -> None:
27        """Load the test set users that all models can use for testing.
28
29        Predictions are made for every user-item pair in the test set.
30
31        Raises:
32            FileNotFoundError: when the test set file is not found.
33        """
34        self.test_set_users = self.load_test_set_dataframe('prediction test set')
35
36    def test_model_ratings(
37            self,
38            model: BasePredictor,
39            user_batch: List[int],
40            **kwargs) -> pd.DataFrame:
41        """Test the specified model for rating predictions.
42
43        Predict ratings for each user-item pair that is present in the test set.
44
45        Args:
46            model: the model that needs to be tested.
47            user_batch: the user batch to compute model ratings for.
48
49        Raises:
50            ArithmeticError: possibly raised by a predictor model on testing.
51            MemoryError: possibly raised by a predictor model on testing.
52            RuntimeError: possibly raised by a predictor model on testing.
53
54        Returns:
55            a dataframe containing the computed rating predictions.
56        """
57        return model.predict_batch(user_batch)

class PredictionPipeline(src.fairreckitlib.model.pipeline.model_pipeline.ModelPipeline): View Source

21class PredictionPipeline(ModelPipeline):
22    """Prediction Pipeline that computes user/item rating predictions.
23
24    The (user,item) prediction will be computed and for each pair that is present in the test set.
25    """
26
27    def load_test_set_users(self) -> None:
28        """Load the test set users that all models can use for testing.
29
30        Predictions are made for every user-item pair in the test set.
31
32        Raises:
33            FileNotFoundError: when the test set file is not found.
34        """
35        self.test_set_users = self.load_test_set_dataframe('prediction test set')
36
37    def test_model_ratings(
38            self,
39            model: BasePredictor,
40            user_batch: List[int],
41            **kwargs) -> pd.DataFrame:
42        """Test the specified model for rating predictions.
43
44        Predict ratings for each user-item pair that is present in the test set.
45
46        Args:
47            model: the model that needs to be tested.
48            user_batch: the user batch to compute model ratings for.
49
50        Raises:
51            ArithmeticError: possibly raised by a predictor model on testing.
52            MemoryError: possibly raised by a predictor model on testing.
53            RuntimeError: possibly raised by a predictor model on testing.
54
55        Returns:
56            a dataframe containing the computed rating predictions.
57        """
58        return model.predict_batch(user_batch)

Prediction Pipeline that computes user/item rating predictions.

The (user,item) prediction will be computed and for each pair that is present in the test set.

def load_test_set_users(self) -> None: View Source

27    def load_test_set_users(self) -> None:
28        """Load the test set users that all models can use for testing.
29
30        Predictions are made for every user-item pair in the test set.
31
32        Raises:
33            FileNotFoundError: when the test set file is not found.
34        """
35        self.test_set_users = self.load_test_set_dataframe('prediction test set')

Load the test set users that all models can use for testing.

Predictions are made for every user-item pair in the test set.

Raises: FileNotFoundError: when the test set file is not found.

def test_model_ratings( self, model: src.fairreckitlib.model.algorithms.base_predictor.BasePredictor, user_batch: List[int], **kwargs) -> pandas.core.frame.DataFrame: View Source

37    def test_model_ratings(
38            self,
39            model: BasePredictor,
40            user_batch: List[int],
41            **kwargs) -> pd.DataFrame:
42        """Test the specified model for rating predictions.
43
44        Predict ratings for each user-item pair that is present in the test set.
45
46        Args:
47            model: the model that needs to be tested.
48            user_batch: the user batch to compute model ratings for.
49
50        Raises:
51            ArithmeticError: possibly raised by a predictor model on testing.
52            MemoryError: possibly raised by a predictor model on testing.
53            RuntimeError: possibly raised by a predictor model on testing.
54
55        Returns:
56            a dataframe containing the computed rating predictions.
57        """
58        return model.predict_batch(user_batch)

Test the specified model for rating predictions.

Predict ratings for each user-item pair that is present in the test set.

Args: model: the model that needs to be tested. user_batch: the user batch to compute model ratings for.

Raises: ArithmeticError: possibly raised by a predictor model on testing. MemoryError: possibly raised by a predictor model on testing. RuntimeError: possibly raised by a predictor model on testing.

Returns: a dataframe containing the computed rating predictions.

Inherited Members

src.fairreckitlib.model.pipeline.model_pipeline.ModelPipeline: ModelPipeline; run; run_model; begin_model; create_model_output_dir; get_model_output_dir; end_model; on_load_train_set_matrix; load_train_set_matrix; load_train_set_dataframe; load_test_set_dataframe; reconstruct_ratings; test_model; train_model; train_and_test_model
src.fairreckitlib.core.pipeline.core_pipeline.CorePipeline: read_dataframe; write_dataframe