Update all baseline models.

2026-07-07 13:00:58 +08:00 · 2020-11-27 22:30:05 +08:00
parent 7952d79932
commit bebce24a7c
17 changed files with 282 additions and 856 deletions
--- a/examples/benchmarks/GATs/workflow_config_gats.yaml
+++ b/examples/benchmarks/GATs/workflow_config_gats.yaml
@@ -8,6 +8,20 @@ data_handler_config: &data_handler_config
    fit_start_time: 2008-01-01
    fit_end_time: 2014-12-31
    instruments: *market
+    infer_processors:
+        - class: RobustZScoreNorm
+          kwargs:
+              fields_group: feature
+              clip_outlier: true
+        - class: Fillna
+          kwargs:
+              fields_group: feature
+    learn_processors:
+        - class: DropnaLabel
+        - class: CSRankNorm
+          kwargs:
+              fields_group: label
+    label: ["Ref($close, -2) / Ref($close, -1) - 1"]
 port_analysis_config: &port_analysis_config
    strategy:
        class: TopkDropoutStrategy
@@ -26,8 +40,8 @@ port_analysis_config: &port_analysis_config
        min_cost: 5
 task:
    model:
-        class: GAT
-        module_path: qlib.contrib.model.pytorch_gats
+        class: GAT_Classic
+        module_path: qlib.contrib.model.pytorch_gats_classic
        kwargs:
            d_feat: 6
            hidden_size: 64
@@ -38,8 +52,7 @@ task:
            early_stop: 20
            metric: loss
            loss: mse
-            base_model: LSTM
-            with_pretrain: True
+            base_model: GRU
            seed: 0
            GPU: 0
    dataset:
@@ -47,7 +60,7 @@ task:
        module_path: qlib.data.dataset
        kwargs:
            handler:
-                class: ALPHA360_Denoise
+                class: ALPHA360
                module_path: qlib.contrib.data.handler
                kwargs: *data_handler_config
            segments:
@@ -58,11 +71,6 @@ task:
        - class: SignalRecord
          module_path: qlib.workflow.record_temp
          kwargs: {}
-        - class: SigAnaRecord
-          module_path: qlib.workflow.record_temp
-          kwargs: 
-            ana_long_short: False
-            ann_scaler: 252
        - class: PortAnaRecord
          module_path: qlib.workflow.record_temp
          kwargs: 
--- a/examples/benchmarks/HATS/README.md
+++ b/examples/benchmarks/HATS/README.md
@@ -1,15 +0,0 @@
-## Requirement
-
-* pandas==1.1.2
-* numpy==1.17.4
-* scikit_learn==0.23.2
-* torch==1.7.0
-
-## HATS
-
-* HATS is a a hierarchical attention network for stock prediction which uses relational data for stock market prediction. HATS selectively aggregates information
-on different relation types and adds the information to the representations of each company. HATS is used as a relational modeling module with initialized node representations.Furthermore, HATS
-can  predict not only individual stock prices but also market index movements, which is similar to the graph classification task.
-
-* HATS uses pretrained model of GRU and LSTM. The code of GRU and LSTM used in Qlib is a pyTorch implemention of GRU and LSTM.
-* Paper address:HATS: A Hierarchical Graph Attention Network for Stock Movement Prediction https://arxiv.org/pdf/1908.07999.pdf
--- a/examples/benchmarks/HATS/requirements.txt
+++ b/examples/benchmarks/HATS/requirements.txt
@@ -1,4 +0,0 @@
-pandas==1.1.2
-numpy==1.17.4
-scikit_learn==0.23.2
-torch==1.7.0
--- a/examples/benchmarks/HATS/worflow_config_hats.yaml
+++ b/examples/benchmarks/HATS/worflow_config_hats.yaml
@@ -1,77 +0,0 @@
-provider_uri: "~/.qlib/qlib_data/cn_data"
-region: cn
-market: &market csi300
-benchmark: &benchmark SH000300
-data_handler_config: &data_handler_config
-    start_time: 2008-01-01
-    end_time: 2020-08-01
-    fit_start_time: 2008-01-01
-    fit_end_time: 2014-12-31
-    instruments: *market
-    infer_processors:
-        - class: RobustZScoreNorm
-          kwargs:
-              fields_group: feature
-              clip_outlier: true
-        - class: Fillna
-          kwargs:
-              fields_group: feature
-    learn_processors:
-        - class: DropnaLabel
-        - class: CSRankNorm
-          kwargs:
-              fields_group: label
-    label: ["Ref($close, -2) / Ref($close, -1) - 1"]
-port_analysis_config: &port_analysis_config
-    strategy:
-        class: TopkDropoutStrategy
-        module_path: qlib.contrib.strategy.strategy
-        kwargs:
-            topk: 50
-            n_drop: 5
-    backtest:
-        verbose: False
-        limit_threshold: 0.095
-        account: 100000000
-        benchmark: *benchmark
-        deal_price: close
-        open_cost: 0.0005
-        close_cost: 0.0015
-        min_cost: 5
-task:
-    model:
-        class: HATS
-        module_path: qlib.contrib.model.pytorch_hats
-        kwargs:
-            d_feat: 6
-            hidden_size: 64
-            num_layers: 2
-            dropout: 0.6
-            n_epochs: 200
-            lr: 1e-3
-            early_stop: 20
-            metric: loss
-            loss: mse
-            base_model: GRU
-            seed: 0
-            GPU: 0
-    dataset:
-        class: DatasetH
-        module_path: qlib.data.dataset
-        kwargs:
-            handler:
-                class: ALPHA360
-                module_path: qlib.contrib.data.handler
-                kwargs: *data_handler_config
-            segments:
-                train: [2008-01-01, 2014-12-31]
-                valid: [2015-01-01, 2016-12-31]
-                test: [2017-01-01, 2020-08-01]
-    record: 
-        - class: SignalRecord
-          module_path: qlib.workflow.record_temp
-          kwargs: {}
-        - class: PortAnaRecord
-          module_path: qlib.workflow.record_temp
-          kwargs: 
-            config: *port_analysis_config
--- a/examples/benchmarks/LSTM/model_lstm_csi300.pkl
+++ b/examples/benchmarks/LSTM/model_lstm_csi300.pkl
--- a/examples/benchmarks/TabNet/README.md
+++ b/examples/benchmarks/TabNet/README.md
@@ -1,4 +0,0 @@
-# TabNet
-* TabNet is a novel high-performance and interpretable canonical deep tabular data learning architectur. TabNet uses sequential attention to choose which features to reason from at each decision step, enabling interpretability and more effcient learning as the learning capacity is used for the most salient features.
-* The code used in Qlib is a pyTorch implementation of Tabnet (Arik, S. O., & Pfister, T. (2019). [https://github.com/dreamquark-ai/tabnet](https://github.com/dreamquark-ai/tabnet)
-* Paper: TabNet: Attentive Interpretable Tabular Learning. [https://arxiv.org/pdf/1908.07442.pdf](https://arxiv.org/pdf/1908.07442.pdf).
--- a/examples/benchmarks/TabNet/requirements.txt
+++ b/examples/benchmarks/TabNet/requirements.txt
@@ -1,5 +0,0 @@
-pandas==1.1.2
-numpy==1.17.4
-scikit_learn==0.23.2
-torch==1.7.0
-pytorch-tabnet==2.0.1
--- a/examples/benchmarks/TabNet/workflow_config_tabnet.yaml
+++ b/examples/benchmarks/TabNet/workflow_config_tabnet.yaml
@@ -1,66 +0,0 @@
-provider_uri: "~/.qlib/qlib_data/cn_data"
-region: cn
-market: &market csi300
-benchmark: &benchmark SH000300
-data_handler_config: &data_handler_config
-    start_time: 2008-01-01
-    end_time: 2020-08-01
-    fit_start_time: 2008-01-01
-    fit_end_time: 2014-12-31
-    instruments: *market
-port_analysis_config: &port_analysis_config
-    strategy:
-        class: TopkDropoutStrategy
-        module_path: qlib.contrib.strategy.strategy
-        kwargs:
-            topk: 50
-            n_drop: 5
-    backtest:
-        verbose: False
-        limit_threshold: 0.095
-        account: 100000000
-        benchmark: *benchmark
-        deal_price: close
-        open_cost: 0.0005
-        close_cost: 0.0015
-        min_cost: 5
-task:
-    model:
-        class: TabNetModel
-        module_path: qlib.contrib.model.tabnet
-        kwargs:
-            n_d: 8
-            n_a: 8
-            n_steps: 3
-            gamma: 1.3
-            n_independent: 2
-            n_shared: 2
-            seed: 0
-            momentum: 0.02
-            lambda_sparse: 1e-3
-            optimizer_params: {lr: 2e-3}
-    dataset:
-        class: DatasetH
-        module_path: qlib.data.dataset
-        kwargs:
-            handler:
-                class: ALPHA360_Denoise
-                module_path: qlib.contrib.data.handler
-                kwargs: *data_handler_config
-            segments:
-                train: [2008-01-01, 2014-12-31]
-                valid: [2015-01-01, 2016-12-31]
-                test: [2017-01-01, 2020-08-01]
-    record: 
-        - class: SignalRecord
-          module_path: qlib.workflow.record_temp
-          kwargs: {}
-        - class: SigAnaRecord
-          module_path: qlib.workflow.record_temp
-          kwargs: 
-            ana_long_short: False
-            ann_scaler: 252
-        - class: PortAnaRecord
-          module_path: qlib.workflow.record_temp
-          kwargs: 
-            config: *port_analysis_config