def test_gist_case(self):
# https://gist.github.com/bwhite/3726239#gistcomment-2852580
res = recommender_metrics.calculate_metrics(
k_list=[1, 2, 3],
group_ids=np.asarray([1, 1, 1, 2, 2, 2, 3, 3, 3]),
scores=np.asarray([1, 2, 3, 1, 2, 3, 1, 2, 3]),
labels=np.asarray([1, 1, 1, 0, 1, 1, 0, 0, 1]),
reduce=False,
)
target = {
"mAP@1": np.asarray([1.0, 1.0, 1.0]),
"precision@1": np.asarray([1.0, 1.0, 1.0]),
"recall@1": np.asarray([0.33333333, 0.5, 1.0]),
"mAP@2": np.asarray([1.0, 1.0, 1.0]),
"precision@2": np.asarray([1.0, 1.0, 0.5]),
"recall@2": np.asarray([0.66666667, 1.0, 1.0]),
"mAP@3": np.asarray([1.0, 1.0, 1.0]),
"precision@3": np.asarray([1.0, 0.66666667, 0.33333333]),
"recall@3": np.asarray([1.0, 1.0, 1.0]),
}
for key, val in target.items():
assert np.allclose(val, res[key])
def basic_usage_example_1():
from recommender_metrics import calculate_metrics
import numpy as np
import json
print("Running example1")
rng = np.random.RandomState(1234)
metrics = calculate_metrics(
group_ids=rng.randint(0, 10, 100),
scores=rng.normal(0, 1, 100),
labels=rng.rand(100) > 0.8,
)
print(json.dumps(metrics, indent=2))
print("\n\n\n")
def basic_usage_large_dataset():
from recommender_metrics import calculate_metrics
from recommender_metrics import generate_random_data
import json
groups, scores, labels = generate_random_data(n_users=50000)
print("Larger data:")
print(" #groups:", len(groups))
print(" #scores:", len(scores))
print(" #labels:", len(labels))
print()
metrics = calculate_metrics(group_ids=groups, scores=scores, labels=labels)
print("Metrics:")
print(json.dumps(metrics, indent=2))
print("\n\n\n")
def basic_usage_group_filttering():
from recommender_metrics import calculate_metrics
from recommender_metrics import generate_random_data
import json
print("Running example3")
groups, scores, labels = generate_random_data()
print("Data:")
print(" #groups:", len(groups))
print(" #scores:", len(scores))
print(" #labels:", len(labels))
print()
metrics = calculate_metrics(group_ids=groups,
scores=scores,
labels=labels,
remove_empty=True)
print("Metrics:")
print(json.dumps(metrics, indent=2))
print("\n\n\n")
def basic_usage_ascending_scores():
from recommender_metrics import calculate_metrics
from recommender_metrics import search_data
import json
print("Running example2")
groups, positions, labels = search_data()
print("Data:")
print(" groups:", groups)
print(" positions:", positions)
print(" labels:", labels)
print()
metrics = calculate_metrics(group_ids=groups,
scores=positions,
labels=labels,
ascending=True)
print("Metrics:")
print(json.dumps(metrics, indent=2))
print("\n\n\n")
def basic_usage_custom_metrics_k():
from recommender_metrics import calculate_metrics
from recommender_metrics import generate_random_data
import json
print("Running example4")
groups, scores, labels = generate_random_data()
print("Data:")
print(" #groups:", len(groups))
print(" #scores:", len(scores))
print(" #labels:", len(labels))
print()
metrics = calculate_metrics(
group_ids=groups,
scores=scores,
labels=labels,
k_list=[1, 2, 4, 8, 16],
metrics=["mAP", "precision", "recall", "ndcg", "auroc"],
)
print("Metrics:")
print(json.dumps(metrics, indent=2))
print("\n\n\n")
def test_empty(self):
target = {
"mAP@1": 0.2631578947368421,
"precision@1": 0.2631578947368421,
"recall@1": 0.07631578947368421,
"mAP@5": 0.5046783625730993,
"precision@5": 0.3140350877192983,
"recall@5": 0.45743525480367586,
"mAP@10": 0.4538314536340852,
"precision@10": 0.2941520467836258,
"recall@10": 0.6738512949039264,
"mAP@20": 0.424754158736318,
"precision@20": 0.33355881010883454,
"recall@20": 1.0,
}
groups, scores, labels = recommender_metrics.generate_random_data()
metrics = recommender_metrics.calculate_metrics(group_ids=groups,
scores=scores,
labels=labels,
remove_empty=True)
self.dict_vals_all_close(target=target,
pred=metrics,
desc=f"Removal of empty group labels")