def main():
dsname = "kosarak-jaccard"
# Load the dataset. This assumes you've run the download.sh script.
dataset = open_dataset(os.path.join(ANNB_ROOT, f"{dsname}.hdf5"))
print(
f"Loaded {len(dataset.corpus)} vectors and {len(dataset.queries)} queries"
)
# for _ in range(3):
# loss = exact(dataset)
# print(f"exact: {loss}")
#
# for _ in range(3):
# loss = indexed(dataset)
# print(f"jaccard indexed: {loss}")
#
# for _ in range(3):
# loss = lsh(dataset, 165, 1, 1.5)
# print(f"lsh: {loss}")
bands = [('bands', b) for b in range(10, 601, 10)]
rows = [('rows', r) for r in range(1, 2)]
candidates = [('candidates', c) for c in np.linspace(0, 10, 21)]
combinations = list(map(dict, itertools.product(bands, rows, candidates)))
metrics = np.zeros((len(combinations), 2))
for i, params in enumerate(combinations):
print(f"Running {i + 1} of {len(combinations)}: {params}...")
try:
(x, y) = lsh(dataset, **params)
print(f"Loss = {(x, y)}")
metrics[i] = [x, y]
pmax = pareto_max(metrics)
plt.title(f"{dsname} results")
plt.scatter(metrics[:, 0], metrics[:, 1], label='All')
plt.scatter(metrics[pmax, 0], metrics[pmax, 1], label='Optimal')
plt.legend()
plt.savefig(f"out/{dsname}.png")
plt.clf()
with open(f"out/{dsname}.txt", "w") as fp:
for j in pmax:
d, m = combinations[j], metrics[j]
fp.write(
f"{d['num_bands']}, {d['num_rows']}, {m[0]}, {m[1]}\n")
except Exception as e:
print(e, file=sys.stderr)
continue
finally:
print('-' * 100)
def initialize_dataset(self, data_source):
filepath = utils.get_filepath(data_source, 'train')
self.dataset = utils.open_dataset(filepath)
self.slc_dataset = utils.open_slc_dataset(filepath, self.num_covs)
self.num_classes = self.slc_dataset.num_classes()
def frequency_count_class():
# retrieve class name frequency (top 50)
plt.subplots(figsize=(9, 5))
z = sns.countplot(y="Class Name",
data=df,
order=df["Class Name"].value_counts()[:50].index)
show_values_on_bars(z, "h", 0.3)
plt.title("Frequency Count of Class Name")
plt.xlabel("Count")
plt.show()
if __name__ == '__main__':
df = open_dataset()
# many_trials()
# age_plot()
# department_age_plot()
# counts_department_plot()
# rating_age_plot()
# rating_age_box_plot()
# review_length_plot
# clothing_recommended_boxplot()
# pair_plot_department()
missing_values_plot()
distribution_ratings()
distribution_of_reviews('Division Name', "Reviews in each Division")
distribution_of_reviews('Department Name', "Reviews in each Department")
distribution_of_reviews('Class Name', "Reviews in each Class")
def initialize_dataset(self, data_source):
filepath = utils.get_filepath(data_source, 'train')
self.dataset = utils.open_dataset(filepath)
self.slc_dataset = utils.open_slc_dataset(filepath)
"instance": "36Guadalajara30.txt",
"obj": 57476,
"time": "1.16"
}, {
"instance": "37Guadalajara20.txt",
"obj": 59493,
"time": "2.29"
}]
datasets = [{"instance": "1Bari30.txt", "obj": 14600, "time": "0.06"}]
for dataset in datasets:
print(dataset["instance"])
# read dataset
n, c, q, Q = utils.open_dataset("dataset/" + dataset["instance"])
N = [i for i in range(1, n)]
V = [0] + N
A = [(i, j) for i in V for j in V]
#m = 80
# build initial solution
source = Node(0, q[0])
nodes = [Node(i, q[i]) for i in range(1, n)]
network = Network(source, c, Q)
network.add_nodes(nodes)
routes, total_cost = network.build_route()