Skip to main content

yet another general purpose naive bayesian classifier

Project description

Naive Bayesian Classifier
======================

yet another general purpose Naive Bayesian classifier. (under heavy development)

##Installation
You can install this package using the following ```pip``` command:

```sh
$ sudo pip install naiveBayesClassifier
```


##Example

```python
"""
Suppose you have some texts of news and know their categories.
You want to train a system with this pre-categorized/pre-classified
texts. So, you have better call this data your training set.
"""
from naiveBayesClassifier import tokenizer
from naiveBayesClassifier.trainer import Trainer
from naiveBayesClassifier.classifier import Classifier

newsTrainer = Trainer(tokenizer)

# You need to train the system passing each text one by one to the trainer module.
newsSet =[
{'text': 'not to eat too much is not enough to lose weight', 'category': 'health'},
{'text': 'Russia is trying to invade Ukraine', 'category': 'politics'},
{'text': 'do not neglect exercise', 'category': 'health'},
{'text': 'Syria is the main issue, Obama says', 'category': 'politics'},
{'text': 'eat to lose weight', 'category': 'health'},
{'text': 'you should not eat much', 'category': 'health'}
]

for news in newsSet:
newsTrainer.train(news['text'], news['category'])

# When you have sufficient trained data, you are almost done and can start to use
# a classifier.
newsClassifier = Classifier(newsTrainer.data, tokenizer)

# Now you have a classifier which can give a try to classifiy text of news whose
# category is unknown, yet.
unknownInstance = "Even if I eat too much, is not it possible to lose some weight"
classification = newsClassifier.classify(unknownInstance)

# the classification variable holds the possible categories sorted by
# their probablity value
print classification
```
***Note***: Definitely you will need much more training data than the amount in the above example. Really, a few lines of text like in the example is out of the question to be sufficient training set.



##What is the Naive Bayes Theorem and Classifier
It is needles to explain everything once again here. Instead, one of the most eloquent explanations is quoted here.

The following explanation is quoted from [another Bayes classifier][1] which is written in Go.

> BAYESIAN CLASSIFICATION REFRESHER: suppose you have a set of classes
> (e.g. categories) C := {C_1, ..., C_n}, and a document D consisting
> of words D := {W_1, ..., W_k}. We wish to ascertain the probability
> that the document belongs to some class C_j given some set of
> training data associating documents and classes.
>
> By Bayes' Theorem, we have that
>
> P(C_j|D) = P(D|C_j)*P(C_j)/P(D).
>
> The LHS is the probability that the document belongs to class C_j
> given the document itself (by which is meant, in practice, the word
> frequencies occurring in this document), and our program will
> calculate this probability for each j and spit out the most likely
> class for this document.
>
> P(C_j) is referred to as the "prior" probability, or the probability
> that a document belongs to C_j in general, without seeing the
> document first. P(D|C_j) is the probability of seeing such a
> document, given that it belongs to C_j. Here, by assuming that words
> appear independently in documents (this being the "naive"
> assumption), we can estimate
>
> P(D|C_j) ~= P(W_1|C_j)*...*P(W_k|C_j)
>
> where P(W_i|C_j) is the probability of seeing the given word in a
> document of the given class. Finally, P(D) can be seen as merely a
> scaling factor and is not strictly relevant to classificiation,
> unless you want to normalize the resulting scores and actually see
> probabilities. In this case, note that
>
> P(D) = SUM_j(P(D|C_j)*P(C_j))
>
> One practical issue with performing these calculations is the
> possibility of float64 underflow when calculating P(D|C_j), as
> individual word probabilities can be arbitrarily small, and a
> document can have an arbitrarily large number of them. A typical
> method for dealing with this case is to transform the probability to
> the log domain and perform additions instead of multiplications:
>
> log P(C_j|D) ~ log(P(C_j)) + SUM_i(log P(W_i|C_j))
>
> where i = 1, ..., k. Note that by doing this, we are discarding the
> scaling factor P(D) and our scores are no longer probabilities;
> however, the monotonic relationship of the scores is preserved by the
> log function.

If you are very curious about Naive Bayes Theorem, you may find the following list helpful:

* [Insect Examples][2]
* [Stanford NLP - Bayes Classifier][3]

#Improvements
This classifier uses a very simple tokenizer which is jus a module to split sentences into words. If your training set is large, you can rely on the available tokenizer, otherwise you need to have a better tokenizer specialized to the language of your training texts.

## TODO
* inline docs
* unit-tests

## AUTHORS
* Mustafa Atik @muatik
* Nejdet Yucesoy @nejdetckenobi


[1]: https://github.com/jbrukh/bayesian/blob/master/bayesian.go
[2]: http://www.cs.ucr.edu/~eamonn/CE/Bayesian%20Classification%20withInsect_examples.pdf
[3]: http://nlp.stanford.edu/IR-book/html/htmledition/naive-bayes-text-classification-1.html

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

naiveBayesClassifier-0.1.3.tar.gz (5.2 kB view details)

Uploaded Source

Built Distribution

naiveBayesClassifier-0.1.3-py2.7.egg (10.0 kB view details)

Uploaded Source

File details

Details for the file naiveBayesClassifier-0.1.3.tar.gz.

File metadata

File hashes

Hashes for naiveBayesClassifier-0.1.3.tar.gz
Algorithm Hash digest
SHA256 94fb8082ef9b902c2fb7d2e8c8117d9aa9fb07e10ad43c3908c7235330469cb8
MD5 7207af8df57e7fc70844fdf337bc5102
BLAKE2b-256 79954e1350294827e10d31de03cc1ec166c2eb69e707758447cea848d0513a8d

See more details on using hashes here.

File details

Details for the file naiveBayesClassifier-0.1.3-py2.7.egg.

File metadata

File hashes

Hashes for naiveBayesClassifier-0.1.3-py2.7.egg
Algorithm Hash digest
SHA256 8c7ba5a209ed4b9fdf0da8921478f79c2fbe4dd1ecb38c3768c92c1e99551e3a
MD5 10229b8926c4158faf5264d8545f1da9
BLAKE2b-256 4501efd9aab3d81c8621104b03d5f74e489a4aa3a698d053ece99af649d4ceb7

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page