- remove redundant models #15

- fix docutils to 0.16 so that bullet lists display normally
- update docs #7

docs/index.rst

@@ -9,6 +9,7 @@ Welcome to RePlay's documentation!
    pages/installation
    pages/spark
    pages/modules
+   pages/useful
 
 
 

docs/pages/modules.rst

@@ -13,5 +13,4 @@ Modules
    modules/splitters
    modules/data_preparator
    modules/distributions
-   useful_data/content
 

docs/pages/modules/models.rst

@@ -16,16 +16,13 @@ ___________________
     "K-Nearest Neighbours", "PySpark"
     "Alternating Least Squares", "PySpark"
     "SLIM", "PySpark"
-    "Classifier Recommender", "PySpark"
-    "Stack Recommender", "PySpark"
     "Neural Matrix Factorization", "Python CPU/GPU"
     "MultVAE", "Python CPU/GPU"
     "Word2Vec Recommender", "Python CPU/GPU"
     "ADMM SLIM", "Python CPU"
     "Wilson Recommender", "Python CPU"
     "Обертка Implicit", "Python CPU"
     "Обертка LightFM", "Python CPU"
-    "Stack Recommender", "Depends on base models"
 
 
 To get more info on how to choose base model, please see this  :doc:`page </pages/useful_data/algorithm_selection>`.
@@ -98,22 +95,12 @@ and :math:`w_{ij}\ge 0`
 .. autoclass:: replay.models.SLIM
     :special-members: __init__
 
-Classifier Recommender
-``````````````````````
-..  autoclass:: replay.models.ClassifierRec
-    :special-members: __init__
-
 
 Cluster Recommender
 ```````````````````
 .. autoclass:: replay.models.ClusterRec
    :members:
 
-Stack
-`````
-.. autoclass:: replay.models.Stack
-    :special-members: __init__
-
 
 Neural models with distributed inference
 ________________________________________

docs/pages/useful.rst

@@ -0,0 +1,9 @@
+Useful Info
+============
+
+
+.. toctree::
+   :maxdepth: 2
+   :caption: Contents:
+
+    How to choose a recommender <useful_data/algorithm_selection.md>

docs/pages/useful_data/algorithm_selection.md

@@ -1,28 +1,37 @@
 ## How to choose a recommender
 
 #### Input Data
+
 - _What is the input?_ 
+
 RePlay models differ by types of data they can process:
-    - Collaborative use only user-item interaction logs.
-    - Content-based use only user or item features.
-    - Hybrid can use both log and features.
+
+- Collaborative use only user-item interaction logs.
+- Content-based use only user or item features.
+- Hybrid can use both log and features.
+<br />
+
 - _Are interactions explicit?_
- Our information can be either _explicit_, such as ratings, or _implicit_ such as view, number of play counts. 
+
+ Our information can be either _explicit_ e.g. ratings, or _implicit_ e.g. view, number of play counts. 
 \
 Some models transfrom any type of data to implicit (_unary ratings_).
+
 - _Will there be new users?_
- Some models need to be completely retrained to give predictions for new users while others don't. 
+
+Some models need to be completely retrained to give predictions for new users with few interactions while others don't. 
+
 - _Will there be new items?_
- The same goes for new items.
 
-| Algorithm      | Data         | Interactions | Cold Users | Cold Items |
+The same goes for new items.
+
+| Algorithm      | Data         | Interactions | New Users | New Items |
 | ---------------|--------------|-------|-------|-------|
 |Popular Recommender        |Collaborative    | converted to unary ratings             | + | - |
 |Popular By Users           |Collaborative    | implicit feedback                      | - | - |
 |Wilson Recommender         |Collaborative    | binary ratings                         | + | - |
 |Random Recommender         |Collaborative    | converted to unary ratings             | + | + |
 |K-Nearest Neighbours       |Collaborative    | converted to unary ratings             | + | - |
-|Classifier Recommender      |Content-based    | binary ratings                         | + | + |
 |Alternating Least Squares  |Collaborative    | implicit feedback                      | - | - |
 |Neural Matrix Factorization|Collaborative    | converted to unary ratings             | - | - |
 |SLIM                       |Collaborative    | unary ratings, explicit feedback       | + | - |
@@ -31,15 +40,15 @@ Some models transfrom any type of data to implicit (_unary ratings_).
 |Word2Vec Recommender       |Collaborative    | converted to unary ratings             | + | - |
 |LightFM Wrap               |Hybrid           | [depends on loss](https://making.lyst.com/lightfm/docs/lightfm.html#lightfm)       | + | + |
 |Implicit Wrap              |Collaborative    | [depends on model](https://implicit.readthedocs.io/en/latest/index.html)    | - | - |
-|Stack Recommender          |Hybrid           | `*`  | `*` | `*` |
 |Two Stages Scenario        |Hybrid           | converted to unary ratings for second level    | `*` | `*` |
 
 `*` - depends on base models. 
 
 #### Model requirements
+
 * _Should recommendations be personalized?_ 
-* _Should cold users get recommendations?_ (users without interactions).
-* _Should model recommend cold items?_ (With no interactions).
+* _Should cold users get recommendations?_ (without any interactions).
+* _Should model recommend cold items?_ (that no one interacted with).
 * _Should model be able to recommend unseen items?_
 
 | Algorithm      | Personalized | Cold Users | Cold Items |  Recommends Unseen Items |
@@ -49,7 +58,6 @@ Some models transfrom any type of data to implicit (_unary ratings_).
 |Wilson Recommender           | - | + | - | + |
 |Random Recommender           | - | + | + | + |
 |K-Nearest Neighbours         | + | + | - | + |
-|Classifier Recommender       | + | + | + | + |
 |Alternating Least Squares    | + | - | - | + |
 |Neural Matrix Factorization  | + | - | - | + |
 |SLIM                         | + | - | - | + |
@@ -58,7 +66,6 @@ Some models transfrom any type of data to implicit (_unary ratings_).
 |Word2Vec Recommender         | + | - | - | + |
 |LightFM  Wrap                | + | + | + | + |
 |Implicit Wrap                | + | - | - | + |
-|Stack Recommender            | + | `*` | `*` | `*` |
 |Two Stages Scenario          | + | `*` | `*` | `*` |
 
 `*` - depends on base models. 

pyproject.toml

@@ -69,6 +69,7 @@ pylint = "*"
 Sphinx = "*"
 sphinx-rtd-theme = "*"
 sphinx-autodoc-typehints = "*"
+docutils = 0.16
 recommonmark = "*"
 sphinx-markdown-tables = "*"
 ghp-import = "*"

replay/models/__init__.py

@@ -12,7 +12,6 @@
 from replay.models.association_rules import AssociationRulesItemRec
 from replay.models.base_rec import Recommender
 from replay.models.base_torch_rec import TorchRecommender
-from replay.models.classifier_rec import ClassifierRec
 from replay.models.implicit_wrap import ImplicitWrap
 from replay.models.knn import KNN
 from replay.models.lightfm_wrap import LightFMWrap
@@ -24,5 +23,4 @@
 from replay.models.slim import SLIM
 from replay.models.wilson import Wilson
 from replay.models.word2vec import Word2VecRec
-from replay.models.stack import Stack
 from replay.models.cluster import ClusterRec