nixtlar - v0.6.3

What's Changed

• docs: VN1 Forecasting Competition by @MMenchero in https://github.com/Nixtla/nixtlar/pull/49
• feat: add finetune_depth parameter by @MMenchero in https://github.com/Nixtla/nixtlar/pull/50

Full Changelog: https://github.com/Nixtla/nixtlar/compare/v0.6.2...v0.6.3

- R
Published by MMenchero over 1 year ago

nixtlar - v0.6.2

What's Changed

• hotfix: error raised for short series by @MMenchero in https://github.com/Nixtla/nixtlar/pull/31
• feat: add Azure endpoints by @MMenchero in https://github.com/Nixtla/nixtlar/pull/35
• feat: infer business-day frequency by @MMenchero in https://github.com/Nixtla/nixtlar/pull/36
• fix: issue with Date objects by @MMenchero in https://github.com/Nixtla/nixtlar/pull/37
• docs: add azure vignette by @MMenchero in https://github.com/Nixtla/nixtlar/pull/39

Full Changelog: https://github.com/Nixtla/nixtlar/compare/v0.6.1...v0.6.2

- R
Published by MMenchero over 1 year ago

nixtlar - v0.6.1

What's Changed

• Update NEWS.md by @tracykteal in https://github.com/Nixtla/nixtlar/pull/28
• fix: release v0.6.0 issues by @MMenchero in https://github.com/Nixtla/nixtlar/pull/27
• feat: vectorized date generation by @MMenchero in https://github.com/Nixtla/nixtlar/pull/29

Full Changelog: https://github.com/Nixtla/nixtlar/compare/v0.6.0...v0.6.1

- R
Published by MMenchero over 1 year ago

nixtlar - v0.6.0

v0.6.0

Breaking changes

• The time_col parameter now only accepts values in the formats YYYY-MM-DD and YYYY-MM-DD hh:mm:ss, either as character strings or date objects (Date class or POSIXct class).
• The id_col parameter now defaults to unique_id for all core functions.
• Minimum data requirements have been implemented, following the new TimeGPT API-v2 specifications.
• Changes to the allowed frequency aliases have been made. For more details, see the "Special Topics" vignette.

Features

• Extended support to tibbles for input data.
• Exogenous variables no longer require future values. Users can now provide only the historical values of exogenous variables.

Bug fixes

• Fixed an issue where timestamps such as "2024-01-01 00:00:00" were sometimes converted to "2024-01-01".

Documentation

• Vignettes have been updated to reflect the new features and changes.

Full Changelog: https://github.com/Nixtla/nixtlar/compare/v0.5.4...v0.6.0

- R
Published by MMenchero over 1 year ago

nixtlar - v0.5.4

What's Changed

Version 0.5.4 of nixtlar is now available. This development release introduces parallel processing, which are not yet available on CRAN and can be used to handle large datasets. To use this feature, configure the num_partitions parameter in any of the nixtlar methods as follows:

• num_partitions=NULL: This is the default setting, operating in sequential mode, equivalent to num_partitions=1.
• num_partitions="auto": Automatically uses all available resources as determined by future::availableCores().
• num_partitions can be set to any positive integer. If it exceeds the number of available resources determined by future::availableCores(), it will default to the maximum available cores.

Full Changelog: https://github.com/Nixtla/nixtlar/compare/v0.5.3...v0.5.4

- R
Published by MMenchero almost 2 years ago

nixtlar - nixtlar 0.5.1

- R
Published by MMenchero about 2 years ago