Node2vec (LibCity#235)

* node2vec

* node2vec

contribution_list.md

@@ -72,6 +72,8 @@ For a list of all models reproduced in LibCity, see [Docs](https://bigscity-libc
 ||HMMM|[XBR-1111](https://github.com/XBR-1111)|
 |Road Network Representation Learning|ChebConv|[aptx1231](https://github.com/aptx1231)|
 ||LINE|[l782993610](https://github.com/l782993610)|
+||Node2Vec|[aptx1231](https://github.com/aptx1231)|
+||GAT|[Qwtdgh](https://github.com/Qwtdgh)|
 
 
 

libcity/config/data/RoadNetWorkDataset.json

@@ -0,0 +1,3 @@
+{
+
+}

libcity/config/executor/GensimExecutor.json

@@ -0,0 +1,3 @@
+{
+
+}

libcity/config/model/road_representation/Node2Vec.json

@@ -0,0 +1,11 @@
+{
+    "output_dim": 128,
+    "is_directed": true,
+    "p": 2,
+    "q": 1,
+    "num_walks": 10,
+    "walk_length": 80,
+    "window_size": 10,
+    "num_workers": 10,
+    "max_epoch": 1000
+}

libcity/config/task_config.json

@@ -380,7 +380,7 @@
     },
     "road_representation": {
         "allowed_model": [
-            "ChebConv", "LINE", "GAT"
+            "ChebConv", "LINE", "GAT", "Node2Vec"
         ],
         "allowed_dataset": [
             "BJ_roadmap"
@@ -399,6 +399,11 @@
             "dataset_class": "ChebConvDataset",
             "executor": "ChebConvExecutor",
             "evaluator": "RoadRepresentationEvaluator"
+        },
+        "Node2Vec": {
+            "dataset_class": "RoadNetWorkDataset",
+            "executor": "GensimExecutor",
+            "evaluator": "RoadRepresentationEvaluator"
         }
     }
 }

libcity/data/dataset/__init__.py

@@ -34,6 +34,7 @@
 from libcity.data.dataset.gsnet_dataset import GSNetDataset
 from libcity.data.dataset.line_dataset import LINEDataset
 from libcity.data.dataset.cstn_dataset import CSTNDataset
+from libcity.data.dataset.roadnetwork_dataset import RoadNetWorkDataset
 
 __all__ = [
     "AbstractDataset",
@@ -67,5 +68,6 @@
     'ChebConvDataset',
     "GSNetDataset",
     "LINEDataset",
-    "CSTNDataset"
+    "CSTNDataset",
+    "RoadNetWorkDataset"
 ]

libcity/data/dataset/roadnetwork_dataset.py

@@ -0,0 +1,29 @@
+import os
+from libcity.data.dataset import TrafficStateDataset
+
+
+class RoadNetWorkDataset(TrafficStateDataset):
+    def __init__(self, config):
+        self.config = config
+        self.dataset = self.config.get('dataset', '')
+        self.data_path = './raw_data/' + self.dataset + '/'
+        self.geo_file = self.config.get('geo_file', self.dataset)
+        self.rel_file = self.config.get('rel_file', self.dataset)
+        assert os.path.exists(self.data_path + self.geo_file + '.geo')
+        assert os.path.exists(self.data_path + self.rel_file + '.rel')
+        super().__init__(config)
+
+    def get_data(self):
+        """
+        返回数据的DataLoader，此类只负责返回路网结构adj_mx，而adj_mx在data_feature中，这里什么都不返回
+        """
+        return None, None, None
+
+    def get_data_feature(self):
+        """
+        返回一个 dict，包含数据集的相关特征
+
+        Returns:
+            dict: 包含数据集的相关特征的字典
+        """
+        return {"adj_mx": self.adj_mx, "num_nodes": self.num_nodes}

libcity/evaluator/road_representation_evaluator.py

@@ -46,7 +46,7 @@ def evaluate(self):
         k_means.fit(node_emb)
         y_predict = k_means.predict(node_emb)
 
-        rid_pos = self._load_geo()
+        rid_file = self._load_geo()
         # 记录每个类别都有哪些geo实体
         result_token = dict()
         for i in range(len(y_predict)):
@@ -60,18 +60,27 @@ def evaluate(self):
         self._logger.info('Kmeans category is saved at {}'.format(result_path))
 
         # QGIS可视化
-        rid_pos = rid_pos['coordinates']
+        rid_type = rid_file['type'][0]
+        rid_pos = rid_file['coordinates']
         rid2wkt = dict()
-        for i in range(rid_pos.shape[0]):
-            rid_list = eval(rid_pos[i])
-            wkt_str = 'LINESTRING('
-            for j in range(len(rid_list)):
-                rid = rid_list[j]
-                wkt_str += (str(rid[0]) + ' ' + str(rid[1]))
-                if j != len(rid_list) - 1:
-                    wkt_str += ','
-            wkt_str += ')'
-            rid2wkt[i] = wkt_str
+        if rid_type == 'LineString':
+            for i in range(rid_pos.shape[0]):
+                rid_list = eval(rid_pos[i])  # [(lat1, lon1), (lat2, lon2)...]
+                wkt_str = 'LINESTRING('
+                for j in range(len(rid_list)):
+                    rid = rid_list[j]
+                    wkt_str += (str(rid[0]) + ' ' + str(rid[1]))
+                    if j != len(rid_list) - 1:
+                        wkt_str += ','
+                wkt_str += ')'
+                rid2wkt[i] = wkt_str
+        elif rid_type == 'Point':
+            for i in range(rid_pos.shape[0]):
+                rid_list = eval(rid_pos[i])  # [lat1, lon1]
+                wkt_str = 'Point({} {})'.format(rid_list[0], rid_list[1])
+                rid2wkt[i] = wkt_str
+        else:
+            raise ValueError('Error geo type!')
 
         df = []
         for i in range(len(y_predict)):

libcity/executor/__init__.py

@@ -10,6 +10,7 @@
 from libcity.executor.abstract_tradition_executor import AbstractTraditionExecutor
 from libcity.executor.chebconv_executor import ChebConvExecutor
 from libcity.executor.eta_executor import ETAExecutor
+from libcity.executor.gensim_executor import GensimExecutor
 
 __all__ = [
     "TrajLocPredExecutor",
@@ -23,5 +24,6 @@
     "AbstractTraditionExecutor",
     "ChebConvExecutor",
     "LINEExecutor",
-    "ETAExecutor"
+    "ETAExecutor",
+    "GensimExecutor"
 ]

libcity/executor/gensim_executor.py

@@ -0,0 +1,33 @@
+from libcity.utils import get_evaluator, ensure_dir
+from libcity.executor.abstract_executor import AbstractExecutor
+
+
+class GensimExecutor(AbstractExecutor):
+    def __init__(self, config, model):
+        self.evaluator = get_evaluator(config)
+        self.config = config
+        self.model = model
+        self.exp_id = config.get('exp_id', None)
+
+        self.cache_dir = './libcity/cache/{}/model_cache'.format(self.exp_id)
+        self.evaluate_res_dir = './libcity/cache/{}/evaluate_cache'.format(self.exp_id)
+        ensure_dir(self.cache_dir)
+        ensure_dir(self.evaluate_res_dir)
+
+    def evaluate(self, test_dataloader):
+        """
+        use model to test data
+        """
+        self.evaluator.evaluate()
+
+    def train(self, train_dataloader, eval_dataloader):
+        """
+        use data to train model with config
+        """
+        self.model.run()
+
+    def load_model(self, cache_name):
+        pass
+
+    def save_model(self, cache_name):
+        pass