Skip to content

Releases: tidymodels/embed

embed 1.1.4

20 Mar 18:01
Compare
Choose a tag to compare

Improvements

  • step_umap() has gained initial and target_weight arguments. (#213)

  • Calling ?tidy.step_*() now sends you to the documentation for step_*() where the outcome is documented. (#216)

  • Documentation for tidy methods for all steps has been improved to describe the return value more accurately. (#217)

  • {keras} and {tensorflow} have been moved to Suggests instead of Imports. (#218)

embed 1.1.3

28 Oct 23:46
Compare
Choose a tag to compare
  • step_collapse_stringdist() will now return predictors as factors. (#204)

  • Fixed regression from 1.1.2 in step_lencode_glm() where it couldn't be used on multiple columns.

embed 1.1.2

17 Aug 21:30
Compare
Choose a tag to compare

Improvements

  • The keep_original_cols argument has been added to step_woe(). This change should mean that every step that produces new columns has the keep_original_cols argument. (#194)

  • Many internal changes to improve consistency and slight speed increases.

Breaking Changes

  • step_pca_sparse(), step_pca_truncated() and step_pca_sparse_bayes() now returns data unaltered if num_comp = 0. This is done to be consistent with recipes steps of the same nature. (#190)

embed 1.1.1

05 Jun 19:15
Compare
Choose a tag to compare

Bug Fixes

  • Fixed bug where step_pca_truncated() didn't work with zero selection. (#181)

  • The tidy() methods for step_discretize_cart(), step_discretize_xgb(), step_embed(), step_feature_hash(), step_lencode_bayes(), step_lencode_glm(), step_lencode_mixed(), step_pca_sparse(), step_pca_sparse_bayes(), step_pca_truncated(), step_umap(), and step_woe() now correctly return zero-row tibbles when used with empty selections. (#181)

embed 1.1.0

14 Apr 23:03
Compare
Choose a tag to compare

New Steps

  • step_pca_truncated() has been added. This step only calculates the components that are required, and will be a speedup in cases where it is used on many variables. (#82)

Improvements

  • step_collapse_stringdist() has gained method and options arguments to allow for different types of string distance calculations. (#152)

  • step_umap() has gained the argument metric. (#154)

  • step_embed() has gained the keep_original_cols argument. (#176)

  • All steps now have required_pkgs() methods.

  • Steps with tunable arguments now have those arguments listed in the documentation.

  • All steps that add new columns will now informatively error if name collision occurs.

embed 1.0.0

02 Jul 17:44
Compare
Choose a tag to compare
  • step_collapse_cart() can pool a predictor's factor levels using a tree-based method.

  • step_collapse_stringdist() can pool a predictor's factor levels using string distances.

  • Case weights support have been added to step_discretize_cart(), step_discretize_xgb(), step_lencode_bayes(), step_lencode_glm(), and step_lencode_mixed().

embed 0.2.0

13 Apr 16:43
Compare
Choose a tag to compare
  • step_embed() now correctly defaults to have a random id with the word "embed". (#102)

  • step_feature_hash() is soft deprecated in embed in favor of step_dummy_hash() in textrecipes. (#95)

  • Steps now have a dedicated subsection detailing what happens when tidy() is applied. (#105)

  • Reorganize documentation for all recipe step tidy methods (#115).

  • Fixed a bug where woe_table() and step_woe() didn't respect the factor levels of the outcome. (109)

embed 0.1.5

24 Nov 18:25
1095364
Compare
Choose a tag to compare
  • Re-licensed package from GPL-2 to MIT. See consent from copyright holders here.

  • The tunable parameter ranges for step_umap() were changed for neighbors, num_comp, and min_dist to prevent uwot segmentation faults. The step also check to see if the data dimensions are consistent with the argument values.

  • Two new PCA steps were added, each using sparse techniques for estimation: step_pca_sparse() and step_pca_sparse_bayes().

  • Updated to use recipes_eval_select() from recipes 0.1.17 (#85).

  • Added prefix argument to step_umap() to harmonize with other recipes steps (#93).

  • All embed recipe steps now officially support empty selections to be more aligned with recipes, dplyr and other packages that use tidyselect.

  • step_woe() no longer warns about high-cardinality predictors when the recipe is estimated. Instead it warns when categories have fewer than 10 data points in the training set. (#74)

embed 0.1.4

16 Jan 17:19
Compare
Choose a tag to compare
  • Minor release with changes to test for cases when CRAN cannot get xgboost to work on their Solaris configuration.

  • lme4 and rstanarm are now in the Suggests list so they are not automatically installed with embed. A message is written to the console if those packages are missing and their associated steps functions are invoked.

embed 0.1.3

12 Nov 16:52
Compare
Choose a tag to compare
  • More changes to enable better parallel processing on windows.