Merge pull request #6 from RubixML/0.3.0-beta

0.3.0 beta

.github/FUNDING.yml

@@ -0,0 +1 @@
+github: [RubixML, andrewdalpino]

.github/workflows/ci.yml

@@ -19,6 +19,8 @@ jobs:
         uses: shivammathur/setup-php@v2
         with:
           php-version: ${{ matrix.php-versions }}
+          tools: pecl
+          extensions: bz2
           ini-values: memory_limit=-1    
 
       - name: Validate composer.json

CHANGELOG.md

@@ -1,3 +1,11 @@
+- 0.3.0-beta
+    - Added Vantage Point Tree for spatial queries
+    - Added Bzip2 serializers
+    - Added Levenshtein distance kernel
+    - Move K Best Selector to main repository
+    - Added custom exceptions from the main repo
+    - Moved Flysystem Persister over to main repo
+
 - 0.2.1-beta
     - Implemented K Best feature selector
 

LICENSE

@@ -1,6 +1,6 @@
 MIT License
 
-Copyright (c) 2020 The Rubix ML Community
+Copyright (c) 2020 Rubix ML
 Copyright (c) 2020 Andrew DalPino
 
 Permission is hereby granted, free of charge, to any person obtaining a copy

README.md

@@ -10,5 +10,8 @@ $ composer require rubix/extras
 ### Requirements
 - [PHP](https://php.net/manual/en/install.php) 7.2 or above
 
+##### Optional
+- [Bzip2 extension](https://www.php.net/manual/en/book.bzip2.php) for Bzip2 compression
+
 ## License
-[MIT](https://github.com/RubixML/Extras/blob/master/LICENSE.md)
+The code is licensed [MIT](LICENSE) and the documentation is licensed [CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/).

benchmarks/Transformers/RecursiveFeatureEliminatorBench.php → benchmarks/Graph/Trees/VPTreeBench.php

@@ -1,28 +1,28 @@
 <?php
 
-namespace Rubix\ML\Benchmarks\Transformers;
+namespace Rubix\ML\Benchmarks\Graph\Trees;
 
+use Rubix\ML\Graph\Trees\VPTree;
 use Rubix\ML\Datasets\Generators\Blob;
-use Rubix\ML\Transformers\KBestSelector;
 use Rubix\ML\Datasets\Generators\Agglomerate;
 
 /**
- * @Groups({"Transformers"})
+ * @Groups({"Trees"})
  * @BeforeMethods({"setUp"})
  */
-class KBestSelectorBench
+class VPTreeBench
 {
     protected const DATASET_SIZE = 10000;
 
     /**
-     * @var \Rubix\ML\Datasets\Labeled
+     * @var \Rubix\ML\Datasets\Labeled;
      */
-    public $dataset;
+    protected $dataset;
 
     /**
-     * @var \Rubix\ML\Transformers\KBestSelector
+     * @var \Rubix\ML\Graph\Trees\VPTree
      */
-    protected $transformer;
+    protected $tree;
 
     public function setUp() : void
     {
@@ -34,16 +34,16 @@ public function setUp() : void
 
         $this->dataset = $generator->generate(self::DATASET_SIZE);
 
-        $this->transformer = new KBestSelector(2);
+        $this->tree = new VPTree(30);
     }
 
     /**
      * @Subject
      * @Iterations(3)
-     * @OutputTimeUnit("milliseconds", precision=3)
+     * @OutputTimeUnit("seconds", precision=3)
      */
-    public function apply() : void
+    public function grow() : void
     {
-        $this->dataset->apply($this->transformer);
+        $this->tree->grow($this->dataset);
     }
 }

composer.json

@@ -10,21 +10,24 @@
         "ai", "rubixml", "rubix ml"
     ],
     "authors": [
+        {
+            "name": "Andrew DalPino",
+            "homepage": "https://github.com/andrewdalpino",
+            "role": "Lead Engineer"
+        },
         {
             "name": "Contributors",
             "homepage": "https://github.com/RubixML/Extras/graphs/contributors"
         }
     ],
     "require": {
         "php": ">=7.2",
-        "rubix/ml": "^0.2.0",
+        "rubix/ml": "0.3.0",
         "rubix/tensor": "^2.0.4",
-        "wamania/php-stemmer": "^2.0",
-        "league/flysystem": "2.0.0-beta.3"
+        "wamania/php-stemmer": "^2.0"
     },
     "require-dev": {
         "friendsofphp/php-cs-fixer": "2.16.*",
-        "league/flysystem-memory": "2.0.0-beta.3",
         "phpbench/phpbench": "0.17.*",
         "phpstan/extension-installer": "^1.0",
         "phpstan/phpstan": "0.12.*",

docs/graph/trees/vp-tree.md

@@ -0,0 +1,28 @@
+<span style="float:right;"><a href="https://github.com/RubixML/Extras/blob/master/src/Graph/Trees/VPTree.php">[source]</a></span>
+
+# VP Tree
+A Vantage Point Tree is a binary spatial tree that divides samples by their distance from the center of a cluster called the *vantage point*. Samples that are closer to the vantage point will be put into one branch of the tree while samples that are farther away will be put into the other branch.
+
+**Interfaces:** Binary Tree, Spatial
+
+**Data Type Compatibility:** Depends on distance kernel
+
+## Parameters
+| # | Param | Default | Type | Description |
+|---|---|---|---|---|
+| 1 | max leaf size | 30 | int | The maximum number of samples that each leaf node can contain. |
+| 2 | kernel | Euclidean | Distance | The distance kernel used to compute the distance between sample points. |
+
+## Example
+```php
+use Rubix\ML\Graph\Trees\VPTree;
+use Rubix\ML\Kernels\Distance\Euclidean;
+
+$tree = new VPTree(30, new Euclidean());
+```
+
+## Additional Methods
+This tree does not have any additional methods.
+
+### References
+>- P. N. Yianilos. (1993). Data Structures and Algorithms for Nearest Neighbor Search in General Metric Spaces.

docs/kernels/distance/levenshtein.md

@@ -0,0 +1,16 @@
+<span style="float:right;"><a href="https://github.com/RubixML/Extras/blob/master/src/Kernels/Distance/Levenshtein.php">[source]</a></span>
+
+# Levenshtein
+Levenshtein distance is defined as the number of single-character edits (such as insert, delete, or replace) needed to change one word to another.
+
+**Data Type Compatibility:** Categorical
+
+## Parameters
+This kernel does not have any parameters.
+
+## Example
+```php
+use Rubix\ML\Kernels\Distance\Levenshtein;
+
+$kernel = new Levenshtein();
+```

docs/persisters/Serializers/bzip2.md

@@ -0,0 +1,24 @@
+<span style="float:right;"><a href="https://github.com/RubixML/Extras/blob/master/src/Persisters/Serializers/Bzip2.php">[source]</a></span>
+
+# Bzip2
+A compression format based on the Burrows–Wheeler transform. Bzip2 is slightly smaller than Gzip format but is slower and requires more memory.
+
+> **Note:** This serializer requires the Bzip2 PHP extension.
+
+## Parameters
+| # | Param | Default | Type | Description |
+|---|---|---|---|---|
+| 1 | block size | 4 | int | The size of each block between 1 and 9 where 9 gives the best compression. |
+| 2 | work factor | 0 | int | Controls how the compression phase behaves when the input is highly repetitive. |
+| 3 | serializer | Native | Serializer | The base serializer |
+
+## Example
+```php
+use Rubix\ML\Persisters\Serializers\Bzip2;
+use Rubix\ML\Persisters\Serializers\Native;
+
+$serializer = new Bzip2(4, 125, new Native());
+```
+
+### References
+>- J. Tsai. (2006). Bzip2: Format Specification.

phpunit.xml

@@ -13,12 +13,18 @@
         <testsuite name="Base">
             <directory>tests</directory>
         </testsuite>
+        <testsuite name="Graph">
+            <directory>tests/Graph</directory>
+        </testsuite>
         <testsuite name="NeuralNet">
             <directory>tests/NeuralNet</directory>
         </testsuite>
         <testsuite name="Other">
             <directory>tests/Other</directory>
         </testsuite>
+        <testsuite name="Persisters">
+            <directory>tests/Persisters</directory>
+        </testsuite>
         <testsuite name="Transformers">
             <directory>tests/Transformers</directory>
         </testsuite>