5. PHP-ML use cases
Targeted advertising
Predicting user characteristics: age, sex, language
spoken, interests, occupation and so on
Related content.
Duplicated content.
and thousands of other cases.
14. Support vector regression demo
see:
https://php-ml.readthedocs.io/en/latest/m
achine-learning/regression/svr/
15. PHP ML - Other features
Clustering (K-means algorithm). Similar to SV Classification method. The main
difference of this method is detecting new areas.
Metrics. Provides probability score. We can get the value of chance the some results is
present in the data set.
Pipeline. Allows to use multiple algorithms in the sequence. In other words you can
make calculation in the queue.
Neural network (MLPClassifier). Advanced classification feature. It is much more slower.
Datasets. A base class for handling input data. PHP Array, CSV, TXT and SVM files are
able from the box.
16. PHP-ML Disadvantages.
PHP PERFORMANCE!!!
Arkadiusz Kondas is working on the lib practically alone.
Algorithm c4.5/5.0 (Classification and Regression Tree) is not
implemented. We should use Pipelines + SVR/SVC but it is not very
comfortable.
Some major parameters of algorithms are dropped. For example lift()
in the Apriori