Skip to main content

Dataset Requirements

Data must be organized as time-series data in tabular form, thus have at least two columns, one is timestamp, and other(s) with numerical values.

Input formats supported​

You can create Dataset from:

  • CSV file,
  • SQL-like database.

SQL databases supported:

  • MySQL / MariaDB
  • PostgresSQL
  • Microsoft SQL

Type of values​

Timestamp Numeric (e.g. integer, decimal, float)

Duplicate timestamps​

Your input data can contain multiple records of the same timestamp, TIM Studio will keep the last occurrence only. This "clean up" occurs on Dataset creation, and update.

Time zone​

It is not required to specify time zone in your timestamps. It is expected that your timestamps are in UTC.

Missing values​

If there is data point which value is missing, then leave it blank (empty). DO NOT insert “0” to fill such gaps in data.

Filling the gaps in data is handled by TIM engine automatically by its imputation feature. By default, gaps of max. 6 points are filled automatically, this can be extended to larger gaps in Settings pane in Experiment Workbench (per each Iteration).

Sampling periods​

The sampling period (rate) is detected as the time between subsequent timestamps.

Timestamps can be unevenly spaced.

TIM Studio currently supports the following sampling periods.

Fixed length​

  • 1, 2, 3, 4, 5, 6, 10, 12, 15, 20 or 30 seconds
  • 1, 2, 3, 4, 5, 6, 10, 12, 15, 20 or 30 minutes
  • 1, 2, 3, 4, 6, 8 or 12 hours
  • any number of days

Variable length​

  • any number of months
  • any number of years

Irregular length​

  • from milliseconds to years

Matrix size​

You can upload Dataset with max. size of 4,900,000 data points (e.g. 70,000 rows x 70 columns), the priority is put on rows, afterwards max. allowed number of columns is determined.

For example if you'd want to upload Dataset of 100,000 rows x 50 columns, TIM Studio will first evaluate rows count, that will leave space for 49 columns (4,900,000 / 100,000 = 49), thus you could not upload such Dataset.

Please be aware there is a limit on your TIM license (rows and columns count, i.e. not amount of data points) which is set independently of data upload limit in TIM Studio. E.g. if license limit is set to 50,000 rows, you cannot use aforementioned Dataset with TIM Engine - use it in experiments, or forecast. In such case you'll see warning in Dataset detail page.

We wanted to empower you to make data exploration, obtaining data insights with data that could be potentially bigger than capacity of your license, instead of setting constrain on Dataset upload side, we kept gates open and rather inform you during Dataset exploration.

Data size (bytes)​

Max. size of data that can be uploaded is 50 MB.

Timestamps​

Max. value of timestamp cannot exceed 2262-04-11.

Other​

Each column must have unique (non-empty) name.