fix: bug fixes and input parameters improvements (#76)

* feat: Input parameters entity (#74)

* feat(synth): Change synthesizers input - Add Model and Train parameters as namedTuples.

* feat(examples): Change examples inputs to match changes.

* feat: Change CGAN and WGAN examples.

* feat: Changing GAN parameters.

* feat: Chaing DRAGAN inputs and example file.

* feat(synth): Change learning rate inputs.

* feat: Introduce the concept of epochs.

* fix: Adding stock data dataset.

* feat: Adding test datasets.

* fix: Save and load methods from DRAGAN, WGAN and WGANGP. (#75)

* fix: Save and load methods from DRAGAN, WGAN and WGANGP. (#75)

* fix(TimeGAN):  supervised loss computation (#77)

* removing unused argument in build methods

* Fixed supervisor loss temporal alignment

* temporal allignment fix, path fix on example

* path fix on example ipynb

* docs: add contributing, support and license and minor changes (#78)

* minor changes on quick start and project resources
* add contributing guidelines
* add support and licensing info

* fix: Custom loss function. Example update. (#79)

Co-authored-by: Francisco Santos <36741643+jfsantos-ds@users.noreply.github.com>
Co-authored-by: Arunn Thevapalan <arunn.work@gmail.com>

Author: 3 people

README.md

@@ -21,25 +21,27 @@ Synthetic data can be used for many applications:
 
 # ydata-synthetic
 This repository contains material related with Generative Adversarial Networks for synthetic data generation, in particular regular tabular data and time-series. 
-It consists in a set of different GANs architectures developed ussing Tensorflow 2.0. An example Jupyter Notebook is included, to show how to use the different architectures.
+It consists a set of different GANs architectures developed using Tensorflow 2.0. Several example Jupyter Notebooks and Python scripts are included, to show how to use the different architectures.
 
 # Quickstart
+
+The source code is currently hosted on GitHub at: https://github.com/ydataai/ydata-synthetic
+
+Binary installers for the latest released version are available at the [Python Package Index (PyPI).](https://pypi.org/project/ydata-synthetic/)
 ```
 pip install ydata-synthetic
 ```
 
 ## Examples
 Here you can find usage examples of the package and models to synthesize tabular data.
 
-**Credit Fraud dataset**   [![Open in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/ydataai/ydata-synthetic/blob/master/examples/regular/gan_example.ipynb)
-
-**Stock dataset** [![Open in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/ydataai/ydata-synthetic/blob/master/examples/timeseries/TimeGAN_Synthetic_stock_data.ipynb)
+- Synthesizing the minority class with VanillaGAN on credit fraud dataset  [![Open in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/ydataai/ydata-synthetic/blob/master/examples/regular/gan_example.ipynb)
+- Time Series synthetic data generation with TimeGAN on stock dataset [![Open in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/ydataai/ydata-synthetic/blob/master/examples/timeseries/TimeGAN_Synthetic_stock_data.ipynb)
+- More examples are continously added and can be found in `/examples` directory.
 
 # Project Resources
-- Synthetic GitHub: https://github.com/ydataai/ydata-synthetic
-- Synthetic Data Community Slack: [click here to join](http://slack.ydata.ai/)
 
-### In this repo you can find the following GAN architectures:
+In this repository you can find the several GAN architectures that are used to create synthesizers:
 
 #### Tabular data
 - [GAN](https://arxiv.org/abs/1406.2661)
@@ -51,3 +53,15 @@ Here you can find usage examples of the package and models to synthesize tabular
 
 #### Sequential data
 - [TimeGAN](https://papers.nips.cc/paper/2019/file/c9efe5f26cd17ba6216bbe2a7d26d490-Paper.pdf)
+
+# Contributing
+We are open to collaboration! If you want to start contributing you only need to:
+1. Search for an issue in which you would like to work. Issues for newcomers are labeled with good first issue.
+2. Create a PR solving the issue.
+3. We would review every PRs and either accept or ask for revisions.
+
+# Support
+For support in using this library, please join the #help Slack channel. The Slack community is very friendly and great about quickly answering questions about the use and development of the library. [Click here to join our Slack community!](http://slack.ydata.ai/)
+
+# License
+[GNU General Public License v3.0](https://github.com/ydataai/ydata-synthetic/blob/master/LICENSE)

data/stock_data.csv

@@ -1,5 +1,6 @@
 from ydata_synthetic.preprocessing.regular.adult import transformations
 from ydata_synthetic.synthesizers.regular import DRAGAN
+from ydata_synthetic.synthesizers import ModelParameters, TrainParameters
 
 #Load and process the data
 data, processed_data, preprocessor = transformations()
@@ -12,16 +13,22 @@
 batch_size = 500
 
 log_step = 100
-epochs = 200+1
+epochs = 300+1
 learning_rate = 1e-5
 beta_1 = 0.5
 beta_2 = 0.9
 models_dir = './cache'
 
-gan_args = [batch_size, learning_rate, beta_1, beta_2, noise_dim, processed_data.shape[1], dim]
-train_args = ['', epochs, log_step]
+gan_args = ModelParameters(batch_size=batch_size,
+                           lr=learning_rate,
+                           betas=(beta_1, beta_2),
+                           noise_dim=noise_dim,
+                           n_cols=processed_data.shape[1],
+                           layers_dim=dim)
+
+train_args = TrainParameters(epochs=epochs,
+                             sample_interval=log_step)
 
 synthesizer = DRAGAN(gan_args, n_discriminator=3)
 synthesizer.train(processed_data, train_args)
-
-synth_data = synthesizer.sample(1000)
+synthesizer.save('adult_synth.pkl')

examples/regular/adult_dragan.py

@@ -1,27 +1,36 @@
 from ydata_synthetic.preprocessing.regular.adult import transformations
 from ydata_synthetic.synthesizers.regular import WGAN_GP
+from ydata_synthetic.synthesizers import ModelParameters, TrainParameters
 
 #Load and process the data
 data, processed_data, preprocessor = transformations()
 
 # WGAN_GP training
-#Defininf the training parameters of WGAN_GP
+#Defining the training parameters of WGAN_GP
 
 noise_dim = 32
 dim = 128
 batch_size = 128
 
 log_step = 100
-epochs = 200+1
-learning_rate = 5e-4
+epochs = 300+1
+learning_rate = [5e-4, 3e-3]
 beta_1 = 0.5
 beta_2 = 0.9
 models_dir = './cache'
 
-gan_args = [batch_size, learning_rate, beta_1, beta_2, noise_dim, processed_data.shape[1], dim]
-train_args = ['', epochs, log_step]
+gan_args = ModelParameters(batch_size=batch_size,
+                           lr=learning_rate,
+                           betas=(beta_1, beta_2),
+                           noise_dim=noise_dim,
+                           n_cols=processed_data.shape[1],
+                           layers_dim=dim)
+
+train_args = TrainParameters(epochs=epochs,
+                             sample_interval=log_step)
 
 synthesizer = WGAN_GP(gan_args, n_critic=2)
 synthesizer.train(processed_data, train_args)
 
 synth_data = synthesizer.sample(1000)
+synthesizer.save('test.pkl')

examples/regular/adult_wgangp.py

@@ -1,5 +1,6 @@
 from ydata_synthetic.synthesizers.regular import CGAN
 from ydata_synthetic.preprocessing.regular.credit_fraud import transformations
+from ydata_synthetic.synthesizers import ModelParameters, TrainParameters
 
 import pandas as pd
 import numpy as np
@@ -46,7 +47,7 @@
 beta_2 = 0.9
 
 log_step = 100
-epochs = 500 + 1
+epochs = 300 + 1
 learning_rate = 5e-4
 models_dir = './cache'
 
@@ -57,14 +58,25 @@
 train_sample[ data_cols ] = train_sample[ data_cols ] / 10 # scale to random noise size, one less thing to learn
 train_no_label = train_sample[ data_cols ]
 
-gan_args = [batch_size, learning_rate, beta_1, beta_2, noise_dim, train_sample.shape[1], dim]
-train_args = ['', -1, epochs, log_step, (0, 1)]
+#Test here the new inputs
+gan_args = ModelParameters(batch_size=batch_size,
+                           lr=learning_rate,
+                           betas=(beta_1, beta_2),
+                           noise_dim=noise_dim,
+                           n_cols=train_sample.shape[1],
+                           layers_dim=dim)
+
+train_args = TrainParameters(epochs=epochs,
+                             cache_prefix='',
+                             sample_interval=log_step,
+                             label_dim=-1,
+                             labels=(0,1))
 
 #Init the Conditional GAN providing the index of the label column as one of the arguments
-synthesizer = CGAN(gan_args, num_classes=2)
+synthesizer = CGAN(model_parameters=gan_args, num_classes=2)
 
 #Training the Conditional GAN
-synthesizer.train(train_sample, train_args)
+synthesizer.train(data=train_sample, label="Class",train_arguments=train_args)
 
 #Saving the synthesizer
 synthesizer.save('cgan_synthtrained.pkl')