cat-mod 0.1.0

No project description provided

CatMod (Categorical classification Model)

Short introduction

• In a modern word that required fast robust and simple development process with Machine Learning, AI and Deep Learning, there are countless of projects require Natural Language Processing (NLP) classification problems such as Commodity Classification, Company Type Classification, Food Type classification, etc.
  More and more people want to train, test and deploy NLP classification model without having to know the background of advanced in programming and AI knowledge.
• This Framework will allow everyone to train, test and save and load their own model and deploy it wherever they want with some simple lines of code.

Virtual Enviroment / Dependencies

• It is recommended to create a vitual enviroment for your project when using CatMod as it will download and install packages and dependencies that might conflict with your dependencies on your machine.
• If you don't mind about the version of the libraries listed in the requirements.txt you can leave it as it is.

How to use

• Please clone or download this repository to your project folder, please assure that catmod.py file stay in the same folder of your python main file (The file in which you want to import CatMod)

e.g.

ProjectFolder
|---main.py
|---catmod.py
|---utils
|   |----...
|   |
...

• Download GloVe Embedded Vectors File to the desired foler.

• Import CatMod in your python file.

from cat_mod import CatMod

• Instanciate a new instance with a GloVe Embedded Vectors File.

cat = CatMod('[your_file_path]')

e.g.

file_path = 'C:/User/Desktop/glove.6B.50d.txt'
cat = CatMod(glove_file = file_path)

file_path = 'Machintosh HD/Users/yourName/Desktop/glove.6B.50d.txt'
cat = CatMode(glove_file = file_path)

Training Process

This Framework will allow you to input a .csv file with many columns but you have to specify 2 columns corresponding to values (X) and targets (Y).

Let's say you have a csv file product.csv with columns look like this

company name	product name	category
...	...	...

• You can use 1 out of 2 ways to load the csv file and load the pre-defined model into the instance.

cat.load_csv('[your_csv_file_path]', '[X_column_name]', '[Y_column_name]')
cat.load_model()

e.g.

cat.load_csv('product.csv', 'product name', 'category')
cat.load_model()

OR

cat.load_model('[your_csv_file_path]', '[X_column_name]', '[Y_column_name]')

e.g.

cat.load_model('product.csv', 'product name', 'category')

We can also specified how many LSTM layers you want by adding the corresponding parameter.

cat.load_model('product.csv', 'product name', 'category', num_of_LSTM = 4)

Then we just do one more easy step:

cat.train([number_of_iterations])

e.g.

cat.train(10)

If the number of iterations is not specified the number of iteration is 50. e.g.

cat.train() # 50 iterations

Save Weights

After training you can save your model on your local machine by using .save_weights([name]) method. (No file name suffix is needed)

cat.save_weights('my_model')

If the model is saved successfully we will see the folder appear in the same folder of your project

ProjectFolder
|---main.py
|---catmod.py
|---utils
|   |---...
|   |
|
|---my_model
|   |---...
|   |
|
...

Load Pre-Trained Model

When we have saved the training file, we can reuse it in the future by just loading it back to a new instance.
There are 2 ways of doing it.

The RECOMMENDED way:

from catmod import CatMod

new_cat = CatMod(load_mode = True, load_file = 'my_model')

The other way:

from catmod import CatMod

new_cat = CatMod(glove_file_path = [the_GloVe_file_path_but_it_must_have_the_same_dimension_with_the_pre_trained_model])

new_cat.load_weights('my_model')

Prediction

Prediction the the most easiest and provide many customization so that everyone can predict and export the predict result in .pd, .csv, .xlsx at their own need. e.g.

X = df['X']
new_cat.predict(X, to_csv = True)

The result will export out the csv file that have both column X and Y together.