on_boarding.md

graph LR
    System_Orchestration_Launchers["System Orchestration & Launchers"]
    Configuration_Management["Configuration Management"]
    Data_Management_Preprocessing["Data Management & Preprocessing"]
    Model_Architectures_Baselines_["Model Architectures (Baselines)"]
    Training_Evaluation_Core["Training & Evaluation Core"]
    Optimization_Utilities["Optimization & Utilities"]
    Metrics_Evaluation_Reporting["Metrics & Evaluation Reporting"]
    System_Orchestration_Launchers -- "initiates" --> Configuration_Management
    System_Orchestration_Launchers -- "configures" --> Data_Management_Preprocessing
    System_Orchestration_Launchers -- "configures" --> Model_Architectures_Baselines_
    System_Orchestration_Launchers -- "triggers" --> Training_Evaluation_Core
    Configuration_Management -- "provides configurations to" --> Data_Management_Preprocessing
    Configuration_Management -- "provides configurations to" --> Model_Architectures_Baselines_
    Configuration_Management -- "provides configurations to" --> Training_Evaluation_Core
    Data_Management_Preprocessing -- "provides processed data to" --> Training_Evaluation_Core
    Training_Evaluation_Core -- "executes" --> Model_Architectures_Baselines_
    Training_Evaluation_Core -- "utilizes" --> Optimization_Utilities
    Training_Evaluation_Core -- "sends results to" --> Metrics_Evaluation_Reporting
    Metrics_Evaluation_Reporting -- "provides feedback to" --> System_Orchestration_Launchers
    click Model_Architectures_Baselines_ href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/BasicTS/Model_Architectures_Baselines_.md" "Details"
    click Training_Evaluation_Core href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/BasicTS/Training_Evaluation_Core.md" "Details"
    click Optimization_Utilities href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/BasicTS/Optimization_Utilities.md" "Details"

Loading

Details

The BasicTS project is structured around a configuration-driven architecture, orchestrated by the System Orchestration & Launchers component. This central component initiates and configures various workflows, leveraging the Configuration Management component for all system settings. Data flows from Data Management & Preprocessing, which prepares datasets for the Training & Evaluation Core. The Training & Evaluation Core then interacts with Model Architectures (Baselines) to execute training and evaluation, utilizing Optimization & Utilities for model optimization. Finally, Metrics & Evaluation Reporting processes the results, providing crucial feedback to the System Orchestration & Launchers for iterative refinement and overall system control. This modular design ensures clear separation of concerns, facilitating maintainability and extensibility.

System Orchestration & Launchers

The primary entry points and control mechanisms for initiating various project workflows (training, evaluation, inference). It acts as the orchestrator for the entire system.

Related Classes/Methods:

• experiments/train.py
• basicts/launcher/launcher.py
• server/engine/engine.py

Configuration Management

Centralized handling of all project configurations, enabling flexible and reproducible experiments. This component is crucial for the "Configuration-Driven Architecture" bias.

Related Classes/Methods:

• basicts/utils/config.py

Data Management & Preprocessing

Manages the entire data lifecycle, from raw input to processed datasets ready for model consumption. This includes data loading, transformation, feature engineering, and serialization.

Related Classes/Methods:

• basicts/data/simple_tsf_dataset.py
• basicts/data/simple_inference_dataset.py
• basicts/utils/serialization.py

Model Architectures (Baselines) [Expand]

A comprehensive collection of diverse time series forecasting model implementations, each with its unique architecture. This component represents the core intellectual property of the toolkit.

Related Classes/Methods:

• baselines/Autoformer/arch/autoformer_arch.py

Training & Evaluation Core [Expand]

Encapsulates the fundamental logic for executing training, validation, and testing pipelines. This includes epoch/iteration-based loops, model building, and integration with optimization strategies.

Related Classes/Methods:

• basicts/runners/base_epoch_runner.py
• basicts/runners/base_iteration_runner.py
• basicts/runners/base_tsf_runner.py
• basicts/runners/base_utsf_runner.py
• basicts/runners/runner_zoo/simple_tsf_runner.py

Optimization & Utilities [Expand]

Provides common functionalities for model optimization (optimizers, learning rate schedulers) and general utilities that support the training and evaluation processes.

Related Classes/Methods:

• basicts/runners/optim/builder.py
• basicts/runners/optim/lr_schedulers.py
• basicts/runners/optim/optimizers.py
• basicts/utils/adjacent_matrix_norm.py

Metrics & Evaluation Reporting

Defines and calculates various performance metrics for time series forecasting models and generates comprehensive evaluation reports.

Related Classes/Methods:

• basicts/metrics/mae.py

FAQ