changelog
paulwilcox edited this page Jul 30, 2021
·
5 revisions
- fluent-data.dimReduce presently accepts default eigenvalue <= 1 rule. I would like to change this to do some sort of elbow test instead.
- fluent-data.dimReduce only has one method of reduction and rotation. This can be extended to allow other methods.
- More elegant management of matrix iterative processes under the hood.
- Cluster Analysis
- Convenience methods to matrix for tansform types
- Re-introduce Database Syntax
- Persistence
- Integration with the glimp library
- Video Tutorials
- matrix.get() is now matrix.filter. This is to put distance with dataset.get(), as it is not analagous to that.
- Added dataset.standardize
- Added dataset.regress (wraps fluent-data.regress)
- The regress method now has a log() function for freindly viewing
- Added dataset.dimReduce and fluent-data.dimReduce (factor-analysis type of function)
- Changed it so that fluent-data.round uses a multiplier logic, not a number-of-digits logic
- Added fluent-data.roundToDigits and fluent-data.random
- dataset.log and matrix.log can now accept a number as the third parameter, to round the results more easily than passing in a mapper function (which is still possible).
- dataset.reduce can now work with classical aggregation functions. This may have been possible before, but now it is documented.
- Added dataset.window
- Matrix.pseudoinverse now keeps column names
- Matrix.clone() did not properly clone row/col names. This is fixed.
- Added a bunch of undocumented (mostly statistical) functions to fluent-data. Though at least the documentation points to the fact that they exist.
- Introduced this changemap
- Serious changes and additions in the wiki.
- Changed the way grouping works under the hood.
- Changed the way dataset.log and matrix.log works
- Removed
dataset.with()and 'matrix.logMany`. These can possibly be restored in the future but for now it didn't survive the rebuild. - fluent-data.covMatrix isSample now defaults to false