Download the PHP package assisted-mindfulness/naive-bayes without Composer
On this page you can find all versions of the php package assisted-mindfulness/naive-bayes. It is possible to download/install these versions without Composer. Possible dependencies are resolved automatically.
Download assisted-mindfulness/naive-bayes
More information about assisted-mindfulness/naive-bayes
Files in assisted-mindfulness/naive-bayes
Package naive-bayes
Short Description A fluent Naive Bayes classifier for PHP and Laravel with normalized arbitrary-precision probabilities.
License MIT
Homepage https://github.com/Assisted-Mindfulness/naive-bayes
FAQ
Example:
In general, it is recommended to use always a project to download your libraries. In an application normally there is more than one library needed.
- Some hosting areas are not accessible by a terminal or SSH. Then it is not possible to use Composer.
- To use Composer is sometimes complicated. Especially for beginners.
- Composer needs much resources. Sometimes they are not available on a simple webspace.
- If you are using private repositories you don't need to share your credentials. You can set up everything on our site and then you provide a simple download link to your team member.
- Simplify your Composer build process. Use our own command line tool to download the vendor folder as binary. This makes your build process faster and you don't need to expose your credentials for private repositories.
Informations about the package naive-bayes
Naive Bayes
This PHP package for Naive Bayes works by looking at a training set and making a guess based on that set. It uses simple statistics and a bit of math to calculate the result.
What can I use this for?
You can use this for categorizing any text content into any arbitrary set of categories. For example:
- is an email spam, or not spam ?
- is a news article about technology, politics, or sports ?
- is a piece of text expressing positive emotions, or negative emotions?
Installation
You may install Naive Bayes into your project using the Composer package manager:
Learning
Before the algorithm can do anything, it requires a training set with historical information. To teach your classifier which category the text belongs to, call the learn method:
Categories may be strings, integers, backed enums, or pure enums. Enum cases are normalized to their backing value or, for pure enums, their case name:
The keys returned by guess() and the value returned by most() use the normalized string or integer identifier.
As with native PHP arrays, integer-like string keys such as "10" are normalized to the integer 10.
Guessing
After you have trained the classifier, you can use the prediction of which category the transmitted text belongs to, for example:
To inspect the score for every category, use:
Probabilities are returned as Brick\Math\BigDecimal values and always add up to exactly 1 at the configured decimal
scale. Any rounding remainder is assigned deterministically to the highest-ranked categories. The classifier applies
Laplace smoothing to words learned in other categories and ignores words that do not occur anywhere in the training
data.
Uneven
Categories have equal prior probabilities by default. If the number of learned documents should influence the result, enable uneven mode to calculate Laplace-smoothed priors from the training distribution.
Precision
Returned probabilities use 16 decimal places by default. You can increase or reduce the precision through the fluent
scale method:
The scale must be a positive integer.
Tokenizer
The algorithm utilizes a tokenizer to segment text into words. By default, it extracts Unicode letters, converts them to lowercase, and includes words longer than three characters. You can also define a custom tokenizer:
The tokenizer may return any iterable of strings.
Development
Run the complete fast quality suite:
Run mutation testing with a required MSI of 90%:
The mutation suite measures observable behavior. Protected visibility changes are excluded, while subclass customization is covered by a dedicated behavioral test.
Wrapping up
There you have it! Even with a very small training set the algorithm can still return some decent results. For example, Naive Bayes has been proven to give decent results in sentiment analyses.
Moreover, Naive Bayes can be applied to more than just text. If you have other ways of calculating the probabilities of your metrics, you can plug those in as well.
License
The MIT License (MIT). Please see License File for more information.
