src.fairreckitlib.evaluation.metrics.lenskit.lenskit_rating_metric

This module contains the lenskit rating metric and creation functions.

Classes:

LensKitRatingMetric: rating metric implementation for lenskit.

Functions:

create_mae: create the MAE rating metric (factory creation compatible).
create_rmse: create the RMSE rating metric (factory creation compatible).

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 the lenskit rating metric and creation functions.
 2
 3Classes:
 4
 5    LensKitRatingMetric: rating metric implementation for lenskit.
 6
 7Functions:
 8
 9    create_mae: create the MAE rating metric (factory creation compatible).
10    create_rmse: create the RMSE rating metric (factory creation compatible).
11
12This program has been developed by students from the bachelor Computer Science at
13Utrecht University within the Software Project course.
14© Copyright Utrecht University (Department of Information and Computing Sciences)
15"""
16
17from typing import Any, Dict
18
19from lenskit.metrics import predict
20import pandas as pd
21
22from ...evaluation_sets import EvaluationSets
23from ..metric_base import ColumnMetric
24
25
26class LensKitRatingMetric(ColumnMetric):
27    """Rating metric implementation for the LensKit framework."""
28
29    def on_evaluate(self, eval_sets: EvaluationSets) -> float:
30        """Evaluate the sets for the performance of the metric.
31
32        Args:
33            eval_sets: the sets to use for computing the performance of the metric.
34
35        Returns:
36            the evaluated performance.
37        """
38        lenskit_ratings = eval_sets.ratings.drop('rating', axis=1)
39        score_column = 'score' if 'score' in lenskit_ratings else 'prediction'
40        scores = pd.merge(eval_sets.test, lenskit_ratings, how='left', on=['user', 'item'])
41        return predict.user_metric(scores, score_column=score_column, metric=self.eval_func)
42
43
44def create_mae(name: str, params: Dict[str, Any], **_) -> LensKitRatingMetric:
45    """Create the MAE rating metric.
46
47    Args:
48        name: the name of the metric.
49        params: there are no parameters for this metric.
50
51    Returns:
52        the LensKitAccuracyMetric wrapper of MAE.
53    """
54    return LensKitRatingMetric(name, params, predict.mae)
55
56
57def create_rmse(name: str, params: Dict[str, Any], **_) -> LensKitRatingMetric:
58    """Create the RMSE rating metric.
59
60    Args:
61        name: the name of the metric.
62        params: there are no parameters for this metric.
63
64    Returns:
65        the LensKitAccuracyMetric wrapper of RMSE.
66    """
67    return LensKitRatingMetric(name, params, predict.rmse)

class LensKitRatingMetric(src.fairreckitlib.evaluation.metrics.metric_base.ColumnMetric): View Source

27class LensKitRatingMetric(ColumnMetric):
28    """Rating metric implementation for the LensKit framework."""
29
30    def on_evaluate(self, eval_sets: EvaluationSets) -> float:
31        """Evaluate the sets for the performance of the metric.
32
33        Args:
34            eval_sets: the sets to use for computing the performance of the metric.
35
36        Returns:
37            the evaluated performance.
38        """
39        lenskit_ratings = eval_sets.ratings.drop('rating', axis=1)
40        score_column = 'score' if 'score' in lenskit_ratings else 'prediction'
41        scores = pd.merge(eval_sets.test, lenskit_ratings, how='left', on=['user', 'item'])
42        return predict.user_metric(scores, score_column=score_column, metric=self.eval_func)

Rating metric implementation for the LensKit framework.

def on_evaluate( self, eval_sets: src.fairreckitlib.evaluation.evaluation_sets.EvaluationSets) -> float: View Source

30    def on_evaluate(self, eval_sets: EvaluationSets) -> float:
31        """Evaluate the sets for the performance of the metric.
32
33        Args:
34            eval_sets: the sets to use for computing the performance of the metric.
35
36        Returns:
37            the evaluated performance.
38        """
39        lenskit_ratings = eval_sets.ratings.drop('rating', axis=1)
40        score_column = 'score' if 'score' in lenskit_ratings else 'prediction'
41        scores = pd.merge(eval_sets.test, lenskit_ratings, how='left', on=['user', 'item'])
42        return predict.user_metric(scores, score_column=score_column, metric=self.eval_func)

Evaluate the sets for the performance of the metric.

Args: eval_sets: the sets to use for computing the performance of the metric.

Returns: the evaluated performance.

Inherited Members

src.fairreckitlib.evaluation.metrics.metric_base.ColumnMetric: ColumnMetric
src.fairreckitlib.evaluation.metrics.metric_base.BaseMetric: evaluate; get_name; get_params

def create_mae( name: str, params: Dict[str, Any], **_) -> src.fairreckitlib.evaluation.metrics.lenskit.lenskit_rating_metric.LensKitRatingMetric: View Source

45def create_mae(name: str, params: Dict[str, Any], **_) -> LensKitRatingMetric:
46    """Create the MAE rating metric.
47
48    Args:
49        name: the name of the metric.
50        params: there are no parameters for this metric.
51
52    Returns:
53        the LensKitAccuracyMetric wrapper of MAE.
54    """
55    return LensKitRatingMetric(name, params, predict.mae)

Create the MAE rating metric.

Args: name: the name of the metric. params: there are no parameters for this metric.

Returns: the LensKitAccuracyMetric wrapper of MAE.

def create_rmse( name: str, params: Dict[str, Any], **_) -> src.fairreckitlib.evaluation.metrics.lenskit.lenskit_rating_metric.LensKitRatingMetric: View Source

58def create_rmse(name: str, params: Dict[str, Any], **_) -> LensKitRatingMetric:
59    """Create the RMSE rating metric.
60
61    Args:
62        name: the name of the metric.
63        params: there are no parameters for this metric.
64
65    Returns:
66        the LensKitAccuracyMetric wrapper of RMSE.
67    """
68    return LensKitRatingMetric(name, params, predict.rmse)

Create the RMSE rating metric.

Args: name: the name of the metric. params: there are no parameters for this metric.

Returns: the LensKitAccuracyMetric wrapper of RMSE.