All Versions
26
Latest Version
Avg Release Cycle
16 days
Latest Release
2961 days ago
Changelog History
Page 1
Changelog History
Page 1
-
v0.5.4 Changes
January 06, 2017๐ New Contributors
๐ฅ Breaking Changes
- None
๐ Features
- โ Add a new
datasets
module behind adatasets
feature flag. - โ Add new classification scores:
precision
,recall
, andf1
. - โ Add a new
Transformer::fit
function to allow prefitting of aTransformer
before use.
๐ Bug Fixes
- None
Minor Changes
LinRegressor
now usessolve
instead ofinverse
for improved accuracy and stability.
-
v0.5.3 Changes
December 20, 2016๐ฅ Breaking Changes
- None
๐ Features
- โ Adding a new
confusion_matrix
module.
๐ Bug Fixes
- None
Minor Changes
- โก๏ธ Updated rulinalg dependency to
0.3.7
.
-
v0.5.2 Changes
November 17, 2016๐ New Contributors
๐ฅ Breaking Changes
- None
๐ Features
- None
๐ Bug Fixes
- Regularization constant for GMM is now only added to diagonal.
Minor Changes
- โ Added some better
Result
handling to GMM.
-
v0.5.1 Changes
October 03, 2016This version includes no changes but is a bump due to a crates bug.
๐ See the notes for 0.5.0 below.
-
v0.5.0 Changes
๐ This is another fairly large release. Thank you to everyone who contributed!
๐ New Contributors
๐ฅ Breaking Changes
- The
SupModel
andUnSupModel
traits now returnResult
s for thetrain
andpredict
functions. - โก๏ธ Updated to rulinalg v0.3 (see rulinalg changelog for details).
๐ Features
- โ Adding RMSProp gradient descent algorithm. #121
- โ Adding cross validation. #125
- โ Adding a new
Shuffler
transformer. #135
๐ Bug Fixes
- None
Minor Changes
- โ Adding benchmarks
- Initiate GMM with sample covariance of data (instead of identity matrix).
- The
-
v0.4.4 Changes
August 12, 2016๐ฅ Breaking Changes
- None
๐ Features
- โ Adding new
Transformer
trait for data preprocessing. - โ Adding a
MinMax
transformer. - โ Adding a
Standardizer
transformer.
Minor Changes
- None
-
v0.4.3 Changes
July 28, 2016๐ New Contributors
- ๐ tafia who is responsible for all changes in this release.
๐ฅ Breaking Changes
- None
๐ Features
- None
Minor Changes
- ๐ฏ Made neural nets more efficient by reducing clones and some restructuring.
- ๐ Removing unneeded copying in favour of slicing for performance.
- Using
iter_rows
in favour of manually row iterating by chunks.
-
v0.4.2 Changes
July 24, 2016๐ฅ Breaking Changes
- None
๐ Features
- None
Minor Changes
- ๐ Fixed a significant bug in the K-Means algorithm. Centroids โก๏ธ were not updating correctly during M-step.
-
v0.4.1 Changes
July 21, 2016๐ฅ Breaking Changes
- None
๐ Features
- โ Added experimental implementation of DBSCAN clustering.
Minor Changes
- โ Added new example for K-Means clustering in repo.
-
v0.4.0 Changes
July 12, 2016๐ This is the biggest release so far. Primarily because the
linalg
module has been pulled out into its own crate: rulinalg.In addition to this there have been a number of improvements to the
linalg
๐ andlearning
moduled in this release.๐ฅ Breaking Changes
- The
linalg
module pulled out and replaced by reexports of rulinalg. All structs are now imported at thelinalg
level, i.e.linalg::matrix::Matrix
->linalg::Matrix
. - Decomposition methods now return
Result
instead of panicking on fail. - K-Means now has a trait for
Initializer
- which allows generic initialization algorithms.
๐ Features
- ๐ New error handling in both the
linalg
(now rulinalg) andlearning
modules. - ๐ Bug fixed in eigendecomposition: it can now be used!
- K-means can now take a generic initialization algorithm.
Minor Changes
- Optimization and code cleanup in the decomposition methods.
- Some optimization in the K-Means model.
- The