MatrixFactorizationRecommender

<p>All matrix factorization recommenders inherit from this class</p>

internal

Implements many algorithms for statistical learning on
sparse matrices - matrix factorizations, matrix completion,
elastic net regressions, factorization machines.
Also 'rsparse' enhances 'Matrix' package by providing methods for
multithreaded <sparse, dense> matrix products and native slicing of
the sparse matrices in Compressed Sparse Row (CSR) format.
List of the algorithms for regression problems:
1) Elastic Net regression via Follow The Proximally-Regularized Leader (FTRL)
Stochastic Gradient Descent (SGD), as per McMahan et al(, <doi:10.1145/2487575.2488200>)
2) Factorization Machines via SGD, as per Rendle (2010, <doi:10.1109/ICDM.2010.127>)
List of algorithms for matrix factorization and matrix completion:
1) Weighted Regularized Matrix Factorization (WRMF) via Alternating Least
Squares (ALS) - paper by Hu, Koren, Volinsky (2008, <doi:10.1109/ICDM.2008.22>)
2) Maximum-Margin Matrix Factorization via ALS, paper by Rennie, Srebro
(2005, <doi:10.1145/1102351.1102441>)
3) Fast Truncated Singular Value Decomposition (SVD), Soft-Thresholded SVD,
Soft-Impute matrix completion via ALS - paper by Hastie, Mazumder
et al. (2014, <arXiv:1410.2596>)
4) Linear-Flow matrix factorization, from 'Practical linear models for
large-scale one-class collaborative filtering' by Sedhain, Bui, Kawale et al
(2016, ISBN:978-1-57735-770-4)
5) GlobalVectors (GloVe) matrix factorization via SGD, paper by Pennington,
Socher, Manning (2014, <https://aclanthology.org/D14-1162/>)
Package is reasonably fast and memory efficient - it allows to work with large
datasets - millions of rows and millions of columns. This is particularly useful
for practitioners working on recommender systems.

Dmitriy Selivanov

rsparse

Statistical Learning on Sparse Matrices

David Cortes

Drew Schmidt

Wei-Chen Chen

MatrixFactorizationRecommender function

<p><div class="r6-fields"></div></p><dl>
<dt><code>components</code></dt>
<dd><p>item embeddings</p></dd><p></p><p>
<dt><code>global_bias</code></dt>
<dd><p>global mean (for centering values in explicit feedback)</p></dd></p><p>
<dt><code>global_bias_base</code></dt>
<dd><p>Pre-calculated `-(factors*global_bias)` (for centering values in
implicit feedback when not using user/item biases)</p></dd></p><p>
</p></dl><p></p>

Public fields


<div class="section" id="public-methods">
<h3>Public methods</h3><p></p><p><ul>
<li><p><a href="/link/MatrixFactorizationRecommender%24predict()?package=rsparse&version=0.5.1" data-mini-rdoc="rsparse::MatrixFactorizationRecommender$predict()"><code>MatrixFactorizationRecommender$predict()</code></a></p></li>
<li><p><a href="/link/MatrixFactorizationRecommender%24clone()?package=rsparse&version=0.5.1" data-mini-rdoc="rsparse::MatrixFactorizationRecommender$clone()"><code>MatrixFactorizationRecommender$clone()</code></a></p></li>
</ul></p><p></p></div><p><hr>
<a id="method-MatrixFactorizationRecommender-predict"></a></p><div class="section" id="method-predict-">
<h3>Method <code><a href="/link/predict()?package=rsparse&version=0.5.1" data-mini-rdoc="rsparse::predict()">predict()</a></code></h3>
<p>recommends items for users</p><div class="section" id="usage">
<h4>Usage</h4>
<p><div class="r"></div></p><pre><code>MatrixFactorizationRecommender$predict(
  x,
  k,
  not_recommend = x,
  items_exclude = integer(0),
  ...
)</code></pre><p></p></div><p></p>
</div><p></p><p><div class="section" id="arguments">
<h4>Arguments</h4>
<p><div class="arguments"></div></p><dl>
<dt><code>x</code></dt>
<dd><p>user-item interactions matrix (usually sparse - `Matrix::sparseMatrix`).Users are
rows and items are columns</p></dd></dl></div></p><p>
<dt><code>k</code></dt>
<dd><p>number of items to recommend</p></dd></p><p>
<dt><code>not_recommend</code></dt>
<dd><p>user-item matrix (sparse) which describes which items method should NOT
recomment for each user. Usually this is same as `x` as we don't want to recommend items user
already liked.</p></dd></p><p>
<dt><code>items_exclude</code></dt>
<dd><p>either integer indices or character identifiers of the items to not
recommend to any user.</p></dd></p><p>
<dt><code>...</code></dt>
<dd><p>not used at the moment</p></dd></p><p>
</p><p></p>
<p></p><p></p><p><hr>
<a id="method-MatrixFactorizationRecommender-clone"></a></p><div class="section" id="method-clone-">
<h3>Method <code>clone()</code></h3>
<p>The objects of this class are cloneable with this method.</p><div class="section" id="usage">
<h4>Usage</h4>
<p><div class="r"></div></p><pre><code>MatrixFactorizationRecommender$clone(deep = FALSE)</code></pre><p></p></div><p></p>
</div><p></p><p><div class="section" id="arguments">
<h4>Arguments</h4>
<p><div class="arguments"></div></p><dl>
<dt><code>deep</code></dt>
<dd><p>Whether to make a deep clone.</p></dd></dl></div></p><p>
</p><p></p>
<p></p><p>
</p>

Methods

Base class for matrix factorization recommenders — MatrixFactorizationRecommender

<p><div class="r6-fields"></p><dl>
<dt><code>components</code></dt>
<dd><p>item embeddings</p></dd>


<dt><code>global_bias</code></dt>
<dd><p>global mean (for centering values in explicit feedback)</p></dd>


<dt><code>global_bias_base</code></dt>
<dd><p>Pre-calculated `-(factors*global_bias)` (for centering values in
implicit feedback when not using user/item biases)</p></dd>


</dl><p></div></p>


<div class='section' id='public-methods'>
<h3>Public methods</h3>

<ul>
<li><p><a href='#method-MatrixFactorizationRecommender-predict'><code>MatrixFactorizationRecommender$predict()</code></a></p></li>
<li><p><a href='#method-MatrixFactorizationRecommender-clone'><code>MatrixFactorizationRecommender$clone()</code></a></p></li>
</ul>

</div><p><hr>
<a id="method-MatrixFactorizationRecommender-predict"></a></p><div class='section' id='method-predict-'>
<h3>Method <code><a href='https://rdrr.io/r/stats/predict.html'>predict()</a></code></h3>
<p>recommends items for users</p><div class='section' id='usage'>
<h4>Usage</h4>
<p><div class="r"></p><pre><code>MatrixFactorizationRecommender$predict(
  x,
  k,
  not_recommend = x,
  items_exclude = integer(0),
  ...
)</code></pre><p></div></p>
</div>

<div class='section' id='arguments'>
<h4>Arguments</h4>
<p><div class="arguments"></p><dl>
<dt><code>x</code></dt>
<dd><p>user-item interactions matrix (usually sparse - `Matrix::sparseMatrix`).Users are
rows and items are columns</p></dd>


<dt><code>k</code></dt>
<dd><p>number of items to recommend</p></dd>


<dt><code>not_recommend</code></dt>
<dd><p>user-item matrix (sparse) which describes which items method should NOT
recomment for each user. Usually this is same as `x` as we don't want to recommend items user
already liked.</p></dd>


<dt><code>items_exclude</code></dt>
<dd><p>either integer indices or character identifiers of the items to not
recommend to any user.</p></dd>


<dt><code>...</code></dt>
<dd><p>not used at the moment</p></dd>


</dl><p></div></p>
</div>

</div><p><hr>
<a id="method-MatrixFactorizationRecommender-clone"></a></p><div class='section' id='method-clone-'>
<h3>Method <code>clone()</code></h3>
<p>The objects of this class are cloneable with this method.</p><div class='section' id='usage'>
<h4>Usage</h4>
<p><div class="r"></p><pre><code>MatrixFactorizationRecommender$clone(deep = FALSE)</code></pre><p></div></p>
</div>

<div class='section' id='arguments'>
<h4>Arguments</h4>
<p><div class="arguments"></p><dl>
<dt><code>deep</code></dt>
<dd><p>Whether to make a deep clone.</p></dd>


</dl><p></div></p>
</div>

</div>


MatrixFactorizationRecommender: Base class for matrix factorization recommenders

Description

Arguments

Public fields

Methods

Public methods

Method `predict()`

Usage

Arguments

Method `clone()`

Usage

Arguments

Description

Arguments

Public fields

Methods

Public methods

Method predict()

Usage

Arguments

Method clone()

Usage

Arguments

Method `predict()`

Method `clone()`