JUHE API Marketplace
Back to datasets
Dataset assetOpen Source CommunityMulti‑Domain ApplicationsTime Series Forecasting

Monash-University/monash_tsf

The Monash Time Series Forecasting Repository is the first comprehensive library containing related time‑series datasets, designed to promote global evaluation of forecasting models. All datasets are for research purposes only. The repository includes 30 datasets, comprising publicly available time‑series collections (in various formats) and curated datasets. Many datasets have multiple versions based on frequency and missing‑value handling, resulting in a total of 58 dataset variants. It also includes real‑world and competition time‑series datasets from diverse domains.

Source
hugging_face
Created
Nov 28, 2025
Updated
Jun 13, 2023
Signals
543 views
Availability
Linked source ready
Overview

Dataset description and usage context

Dataset Overview

Basic Information

  • Dataset Name: Monash Time Series Forecasting Repository
  • Dataset Type: Time‑Series Forecasting
  • Language: Single language
  • License: CC‑BY‑4.0
  • Dataset Size: 1K < n < 10K
  • Source Data: Raw data
  • Task Type: Time‑Series Forecasting
  • Task IDs: Univariate time‑series forecasting, Multivariate time‑series forecasting

Dataset Configurations

The dataset comprises multiple configurations, each with distinct features and splits:

Configuration: weather

  • Features:
    • start: timestamp
    • target: float sequence
    • feat_static_cat: unsigned integer sequence
    • feat_dynamic_real: sequence of float sequences
    • item_id: string
  • Splits:
    • train: 3,010 samples, 176,893,738 bytes
    • test: 3,010 samples, 177,638,713 bytes
    • validation: 3,010 samples, 177,266,226 bytes
  • Download Size: 38,820,451 bytes
  • Dataset Size: 531,798,677 bytes

Configuration: tourism_yearly

  • Features: identical to weather
  • Splits:
    • train: 518 samples, 54,264 bytes
    • test: 518 samples, 71,358 bytes
    • validation: 518 samples, 62,811 bytes
  • Download Size: 36,749 bytes
  • Dataset Size: 188,433 bytes

(The remaining configurations follow the same pattern; omitted for brevity.)

Need downstream help?

Pair the dataset with AI analysis and content workflows.

Once the source passes your review, move straight into summarization, transformation, report drafting, or presentation generation with the JuheAI toolchain.

Explore AI studio