Datasets:

THUgewu
/

JapaneseVowels

license: cc-by-4.0
task_categories:
  - time-series-forecasting
tags:
  - time-series
  - forecasting
  - anomaly-detection
  - classification
  - TSLib
  - tsfile
  - modality:timeseries
  - timeseries
  - format:tsfile
pretty_name: JapaneseVowels (TsFile)
size_categories:
  - 1M<n<10M
configs:
  - config_name: default
    data_files:
      - split: train
        path: JapaneseVowels_train.tsfile
      - split: test
        path: JapaneseVowels_test.tsfile
modality:
  - timeseries
  - tabular
  - text
language:
  - en

JapaneseVowels (TsFile)

Apache TsFile version of the JapaneseVowels UEA classification subset of thuml/Time-Series-Library.

Overview

LPC cepstrum of nine male speakers saying /ae/; classify the speaker.

Train samples: 270 • Test samples: 370.
Dimensions (channels): 12 • Series length: 15.
Classes: 9.

Each sample is an independent multivariate series. TRAIN and TEST are stored as two separate TsFiles (JapaneseVowels_train.tsfile / JapaneseVowels_test.tsfile).

Schema (TsFile structure)

sample_index (TAG) — one device per sample. Query one sample with WHERE sample_index=0.
Time (INT64) — within-sample position (0..14); a frame index, not a wall-clock timestamp (the sktime .ts source has none).
dim_0 … dim_11 (FIELD, FLOAT) — the 12 channels at each position.
class_label (FIELD, STRING) — the sample's class, repeated on every row.

Converted from the sktime .ts format. Nothing is dropped: every dimension, time point, and label is preserved.

Usage

Read the .tsfile files with the Apache TsFile Java or Python SDK.

Source & license

Original dataset: https://huggingface.co/datasets/thuml/Time-Series-Library (subset JapaneseVowels)
Author / publisher: thuml (Tsinghua University)
Paper: https://arxiv.org/abs/2407.13278
License: CC BY 4.0