support local folder (#511)

* support local folder

* remove unnecessary random

* KaggleScen Subclass

* small fix

* use template for style description

* update default scen to kaggle

rdagent/app/data_science/conf.py

@@ -7,7 +7,7 @@ class DataScienceBasePropSetting(KaggleBasePropSetting):
 
     # Main components
     ## Scen
-    scen: str = "rdagent.scenarios.data_science.scen.DataScienceScen"
+    scen: str = "rdagent.scenarios.data_science.scen.KaggleScen"
     """Scenario class for data mining model"""
 
     ## proposal

rdagent/app/data_science/loop.py

@@ -1,5 +1,6 @@
 from typing import Any
 
+from pathlib import Path
 import fire
 
 from rdagent.app.data_science.conf import DS_RD_SETTING
@@ -111,7 +112,12 @@ def main(path=None, step_n=None, competition=None):
         DS_RD_SETTING.competition = competition
 
     if DS_RD_SETTING.competition:
-        download_data(competition=DS_RD_SETTING.competition, settings=DS_RD_SETTING)
+        if DS_RD_SETTING.scen.endswith("KaggleScen"):
+            download_data(competition=DS_RD_SETTING.competition, settings=DS_RD_SETTING)
+        else:
+            if not Path(f"{DS_RD_SETTING.local_data_path}/{competition}").exists():
+                logger.error(f"Please prepare data for competition {competition} first.")
+                return
     else:
         logger.error("Please specify competition name.")
     if path is None:

rdagent/components/coder/data_science/ensemble/test.py

@@ -8,7 +8,7 @@
 from rdagent.components.coder.data_science.ensemble import EnsembleCoSTEER
 from rdagent.components.coder.data_science.ensemble.exp import EnsembleTask
 from rdagent.scenarios.data_science.experiment.experiment import DSExperiment
-from rdagent.scenarios.data_science.scen import DataScienceScen
+from rdagent.scenarios.data_science.scen import KaggleScen
 
 # Add the competition folder to path
 COMPETITION_PATH = (
@@ -31,7 +31,7 @@ def load_ensemble_spec():
 
 def develop_one_competition(competition: str):
     # Initialize scenario and coder
-    scen = DataScienceScen(competition=competition)
+    scen = KaggleScen(competition=competition)
     ensemble_coder = EnsembleCoSTEER(scen)
     # Load ensemble specification
     ensemble_spec = load_ensemble_spec()

rdagent/components/coder/data_science/feature/test.py

@@ -9,11 +9,11 @@
 from rdagent.components.coder.data_science.feature import FeatureCoSTEER
 from rdagent.components.coder.data_science.feature.exp import FeatureTask
 from rdagent.scenarios.data_science.experiment.experiment import DSExperiment
-from rdagent.scenarios.data_science.scen import DataScienceScen
+from rdagent.scenarios.data_science.scen import KaggleScen
 
 
 def develop_one_competition(competition: str):  # -> experiment
-    scen = DataScienceScen(competition=competition)
+    scen = KaggleScen(competition=competition)
     feature_coder = FeatureCoSTEER(scen)
 
     with open("./rdagent/scenarios/kaggle/tpl_ex/aerial-cactus-identification/spec/feature.md", "r") as file:

rdagent/components/coder/data_science/model/test.py

@@ -12,12 +12,12 @@
 from rdagent.components.coder.data_science.model.exp import ModelTask
 from rdagent.core.experiment import FBWorkspace
 from rdagent.scenarios.data_science.experiment.experiment import DSExperiment
-from rdagent.scenarios.data_science.scen import DataScienceScen
+from rdagent.scenarios.data_science.scen import KaggleScen
 
 
 # Take tasks, spec.md and feat as input, generate a feedback as output
 def develop_one_competition(competition: str):
-    scen = DataScienceScen(competition=competition)
+    scen = KaggleScen(competition=competition)
     model_coder = ModelCoSTEER(scen)
 
     # Create the task

rdagent/components/coder/data_science/raw_data_loader/test.py

@@ -9,11 +9,11 @@
 from rdagent.components.coder.data_science.raw_data_loader import DataLoaderCoSTEER
 from rdagent.components.coder.data_science.raw_data_loader.exp import DataLoaderTask
 from rdagent.scenarios.data_science.experiment.experiment import DSExperiment
-from rdagent.scenarios.data_science.scen import DataScienceScen
+from rdagent.scenarios.data_science.scen import KaggleScen
 
 
 def develop_one_competition(competition: str):  # -> experiment
-    scen = DataScienceScen(competition=competition)
+    scen = KaggleScen(competition=competition)
     data_loader_coder = DataLoaderCoSTEER(scen)
 
     # Create the experiment

rdagent/components/coder/data_science/workflow/test.py

@@ -12,11 +12,11 @@
 from rdagent.components.coder.data_science.workflow.exp import WorkflowTask
 from rdagent.core.experiment import FBWorkspace
 from rdagent.scenarios.data_science.experiment.experiment import DSExperiment
-from rdagent.scenarios.data_science.scen import DataScienceScen
+from rdagent.scenarios.data_science.scen import KaggleScen
 
 
 def develop_one_competition(competition: str):
-    scen = DataScienceScen(competition=competition)
+    scen = KaggleScen(competition=competition)
     workflow_coder = WorkflowCoSTEER(scen)
 
     wt = WorkflowTask(

rdagent/core/experiment.py

@@ -223,7 +223,7 @@ def __init__(
         based_experiments: Sequence[ASpecificWSForExperiment] = [],
         hypothesis: Optional["Hypothesis"] = None,
     ) -> None:
-        self.hypothesis: Optional["Hypothesis"] = hypothesis  # Experiment is opptionally generated by hypothesis
+        self.hypothesis: Optional["Hypothesis"] = hypothesis  # Experiment is optionally generated by hypothesis
         self.sub_tasks: Sequence[ASpecificTask] = sub_tasks
         self.sub_workspace_list: list[ASpecificWSForSubTasks | None] = [None] * len(self.sub_tasks)
         # TODO:

rdagent/scenarios/data_science/scen/__init__.py

@@ -1,3 +1,4 @@
 from .scen import DataScienceScen
+from .kaggle import KaggleScen
 
-__all__ = ["DataScienceScen"]
+__all__ = ["DataScienceScen", "KaggleScen"]

rdagent/scenarios/data_science/scen/kaggle.py

@@ -0,0 +1,35 @@
+import json
+
+from rdagent.app.data_science.conf import DS_RD_SETTING
+from rdagent.core.scenario import Scenario
+from rdagent.oai.llm_utils import APIBackend
+from rdagent.scenarios.data_science.scen import DataScienceScen
+from rdagent.scenarios.kaggle.kaggle_crawler import (
+    crawl_descriptions,
+    leaderboard_scores,
+)
+from rdagent.utils.agent.tpl import T
+
+
+class KaggleScen(DataScienceScen):
+    """Kaggle Scenario
+    It is based on kaggle now.
+        - But it is not use the same interface with previous kaggle version.
+        - Ideally, we should reuse previous kaggle scenario.
+          But we found that too much scenario unrelated code in kaggle scenario and hard to reuse.
+          So we start from a simple one....
+    """
+    def _get_description(self):
+        return crawl_descriptions(self.competition, DS_RD_SETTING.local_data_path)
+
+    def _get_direction(self):
+        leaderboard = leaderboard_scores(self.competition)
+        return "maximize" if float(leaderboard[0]) > float(leaderboard[-1]) else "minimize"
+
+    @property
+    def rich_style_description(self) -> str:
+        return T(".prompts:rich_style_description").r(
+            name="Kaggle",
+            competition=f"[{self.competition}](https://www.kaggle.com/competitions/{self.competition})",
+        )
+

rdagent/scenarios/data_science/scen/prompts.yaml

@@ -34,4 +34,29 @@ competition_background: |-
   The data type used in this competition is {{ data_type }}.
   Briefly, the competition involves: {{ brief_description }}.
   The dataset used in this competition is: {{ data_description }}.
-  Your goal in this competition is to: {{target_description }}.
+  Your goal in this competition is to: {{target_description }}.
+
+rich_style_description: |-
+  ### {{ name }} Agent: Automated Feature Engineering & Model Tuning Evolution
+
+  #### [Overview](#_summary)
+
+  In this scenario, our automated system proposes hypothesis, choose action, implements code, conducts validation, and utilizes feedback in a continuous, iterative process.
+
+  #### {{ name }} Competition info
+
+  Current Competition: {{ competition }}
+
+  #### [Automated R&D](#_rdloops)
+
+  - **[R (Research)](#_research)**
+  - Iteration of ideas and hypotheses.
+  - Continuous learning and knowledge construction.
+
+  - **[D (Development)](#_development)**
+  - Evolving code generation, model refinement, and features generation.
+  - Automated implementation and testing of models/features.
+
+  #### [Objective](#_summary)
+
+  To automatically optimize performance metrics within the validation set, ultimately discovering the most efficient features and models through autonomous research and development.

rdagent/scenarios/data_science/scen/scen.py

@@ -1,32 +1,33 @@
 import json
+from pathlib import Path
 
 from rdagent.app.data_science.conf import DS_RD_SETTING
 from rdagent.core.scenario import Scenario
 from rdagent.oai.llm_utils import APIBackend
-from rdagent.scenarios.kaggle.kaggle_crawler import (
-    crawl_descriptions,
-    leaderboard_scores,
-)
 from rdagent.utils.agent.tpl import T
+from rdagent.log import rdagent_logger as logger
 
 
 class DataScienceScen(Scenario):
     """Data Science Scenario
-    It is based on kaggle now.
-        - But it is not use the same interface with previous kaggle version.
-        - Ideally, we should reuse previous kaggle scenario.
-          But we found that too much scenario unrelated code in kaggle scenario and hard to reuse.
-          So we start from a simple one....
     """
 
     def __init__(self, competition: str) -> None:
         self.competition = competition
-        self.raw_description = crawl_descriptions(competition, DS_RD_SETTING.local_data_path)
+        self.raw_description = self._get_description()
+        self.metric_direction = self._get_direction()
+        self._analysis_competition_description()
 
-        leaderboard = leaderboard_scores(competition)
-        self.metric_direction = "maximize" if float(leaderboard[0]) > float(leaderboard[-1]) else "minimize"
+    def _get_description(self):
+        if (fp := Path(f"{DS_RD_SETTING.local_data_path}/{self.competition}.json")).exists():
+            logger.info(f"Found {self.competition}.json, loading from local file.")
+            with fp.open("r") as f:
+                return json.load(f)
+        else:
+            logger.error(f"Cannot find {self.competition}.json in {DS_RD_SETTING.local_data_path}, please check the file.")
 
-        self._analysis_competition_description()
+    def _get_direction(self):
+        return self.raw_description.get("metric_direction", "minimize")
 
     def _analysis_competition_description(self):
         sys_prompt = T(".prompts:competition_description_template.system").r()
@@ -75,31 +76,10 @@ def background(self) -> str:
 
     @property
     def rich_style_description(self) -> str:
-        return f"""
-### Kaggle Agent: Automated Feature Engineering & Model Tuning Evolution
-
-#### [Overview](#_summary)
-
-In this scenario, our automated system proposes hypothesis, choose action, implements code, conducts validation, and utilizes feedback in a continuous, iterative process.
-
-#### Kaggle Competition info
-
-Current Competition: [{self.competition}](https://www.kaggle.com/competitions/{self.competition})
-
-#### [Automated R&D](#_rdloops)
-
-- **[R (Research)](#_research)**
-- Iteration of ideas and hypotheses.
-- Continuous learning and knowledge construction.
-
-- **[D (Development)](#_development)**
-- Evolving code generation, model refinement, and features generation.
-- Automated implementation and testing of models/features.
-
-#### [Objective](#_summary)
-
-To automatically optimize performance metrics within the validation set or Kaggle Leaderboard, ultimately discovering the most efficient features and models through autonomous research and development.
-"""
+        return T(".prompts:rich_style_description").r(
+            name="Data Science",
+            competition=self.competition,
+        )
 
     def get_scenario_all_desc(self) -> str:
         return T(".prompts:scenario_description").r(