API Reference#

This page provides detailed documentation for all public classes, functions, and modules in DeepTimeSeries.

The API is organized into the following sections:

Core Modules#

Base classes for building forecasting models. These include ForecastingModule (the foundation for all models), head classes for producing predictions, and metric tracking utilities.

Chunk Specification#

Classes for defining and extracting time windows from time series data. Chunks specify what parts of the data are used for encoding (input), decoding (during prediction), and labels (targets).

Dataset#

The TimeSeriesDataset class provides a PyTorch-compatible dataset for loading time series data with chunk-based extraction.

Data Transformation#

Preprocessing utilities, primarily ColumnTransformer for applying sklearn-style transformers to specific columns of DataFrames.

Models#

Pre-implemented forecasting models including MLP, Dilated CNN, RNN variants (LSTM, GRU), and Transformer. All models support both deterministic and probabilistic forecasting.

Layers#

Custom neural network layers used by the forecasting models, such as positional encoding for transformers and padding layers for causal convolutions.

Utilities#

Helper functions for data manipulation, dictionary operations, and visualization utilities for time series data.

Typical Usage Flow#

  1. Preprocess data using ColumnTransformer

  2. Create a model (e.g., MLP) which automatically generates chunk specifications

  3. Create a dataset using TimeSeriesDataset with the model’s chunk specifications

  4. Train the model using PyTorch Lightning’s Trainer

  5. Use ChunkInverter to convert model outputs back to DataFrames