Click to expand
You can create Machine Learning models from scratch or from templates
Templates
The template is the best approach to begin creating your machine learning model. It allows us to create a machine learning model based on commonly observed problems, for example the MLT.
Available Templates
Actually Tiki only support one template :
More Like This (MLT)
The MLT template solves the problems associated with suggesting similar content (finds documents that are "like" a given set of documents).
This emulates Module More Like This
More info: https://github.com/RubixML/RubixML/issues/75
Transformers and Learners for MoreLikeThis
| Transformers and Applied Learners | Arguments |
|---|---|
| TextNormalizer | |
| StopWordFilter | |
| WordCountVectorizer | maxVocabulary :1000 , minDocumentFrequency :1 ,maxDocumentFrequency: 500 ,okenizer :default |
| BM25Transformer | alpha :1.2 , beta :0.75 |
| KDNeighbors | k:20, weighted:true, tree : BallTree |
Click to expand