libcity.data.dataset.dataset_subclass.geosan_dataset

class libcity.data.dataset.dataset_subclass.geosan_dataset.GeoSANDataset(config)[source]

Bases: libcity.data.dataset.abstract_dataset.AbstractDataset

add_location(loc, coordinate)[source]
build_vocab(min_freq=10)[source]
static collect_fn_quadkey(batch, data_source, sampler, region_processer, loc2quadkey=None, k=5, with_trg_quadkey=True)[source]
get_data()[source]

返回数据的DataLoader,包括训练数据、测试数据(事实上不提供)、验证数据

Returns

tuple contains:

train_dataloader: Dataloader composed of Batch (class)

eval_dataloader: None(no valid step)

test_dataloader: Dataloader composed of Batch (class)

Return type

tuple

get_data_feature()[source]

返回一个 dict,包含数据集的相关特征

Returns

包含数据集的相关特征的字典

Return type

dict

get_visited_locs()[source]
processing(min_freq=20)[source]
region_stats()[source]

统计并打印数据集的一些基本信息

split(max_len=100)[source]
class libcity.data.dataset.dataset_subclass.geosan_dataset.KNNSampler(query_sys, user_visited_locs, num_nearest=100, exclude_visited=False)[source]

Bases: torch.nn.modules.module.Module

forward(trg_seq, k, user, **kwargs)[source]

基于query_sys从候选集中随机采样k个作为负样例

training: bool
class libcity.data.dataset.dataset_subclass.geosan_dataset.LadderSampler(data_source, batch_sz, fix_order=False)[source]

Bases: torch.utils.data.sampler.Sampler

class libcity.data.dataset.dataset_subclass.geosan_dataset.LocQuerySystem[source]

Bases: object

build_tree(dataset)[source]

构建KNN(基于BallTree实现),用于sampler中的采样操作

get_knn(trg_loc, k=100)[source]
get_radius(trg_loc, r=10.0)[source]
prefetch_knn(k=100)[source]
prefetch_radius(radius=10.0)[source]
radius_stats(radius=10)[source]
libcity.data.dataset.dataset_subclass.geosan_dataset.clip(n, min_value, max_value)[source]
libcity.data.dataset.dataset_subclass.geosan_dataset.latlon2pxy(latitude, longitude, level_of_detail)[source]
libcity.data.dataset.dataset_subclass.geosan_dataset.latlon2quadkey(lat, lon, level)[source]

经纬度 to quadkey 转换函数

libcity.data.dataset.dataset_subclass.geosan_dataset.map_size(level_of_detail)[source]
libcity.data.dataset.dataset_subclass.geosan_dataset.pxy2txy(pixel_x, pixel_y)[source]
libcity.data.dataset.dataset_subclass.geosan_dataset.txy2quadkey(tile_x, tile_y, level_of_detail)[source]