Intro
The data organization follows TFB format: https://github.com/decisionintelligence/TFB.
TFB data format
TFB stores time series in a format of three column long tables, which we will introduce below：
Format Introduction
First column: date (the exact column name is required, the same applies below.)
- The columns stores the time information in the time series, which can be in either of the following formats:
- Timestamps in string, datetime, or other types that are compatible with pd.to_datetime;
- Integers starting from 1, e.g. 1, 2, 3, 4, 5, ...
Second column: data
- This column stores the series values corresponding to the timestamps.
Third column: cols
- This column stores the column name (variable name).
Multivariate time series example:
A common time series in wide table format:
| date | channel1 | channel2 | channel3 |
|---|---|---|---|
| 1 | 0.1 | 1 | 10 |
| 2 | 0.2 | 2 | 20 |
| 3 | 0.3 | 3 | 30 |
Convert to TFB format:
| date | data | cols |
|---|---|---|
| 1 | 0.1 | channel1 |
| 2 | 0.2 | channel1 |
| 3 | 0.3 | channel1 |
| 1 | 1 | channel2 |
| 2 | 2 | channel2 |
| 3 | 3 | channel2 |
| 1 | 10 | channel3 |
| 2 | 20 | channel3 |
| 3 | 30 | channel3 |
License
This dataset is released under CC BY 4.0