improved documentation to account for layer config options

Author: maxmekiska

Imbrium.egg-info/PKG-INFO

@@ -15,22 +15,26 @@ Description: # Imbrium
         pip install imbrium
         ```
 
-        Standard and Hybrid Deep Learning Multivariate-Multi-Step
-        (+ Univariate-Multi-Step, because why not?) Time Series Forecasting.
+        Standard and Hybrid Deep Learning Multivariate-Multi-Step & Univariate-Multi-Step
+        Time Series Forecasting.
 
 
-        ██╗███╗░░░███╗██████╗░██████╗░██╗██╗░░░██╗███╗░░░███╗
-        ██║████╗░████║██╔══██╗██╔══██╗██║██║░░░██║████╗░████║
-        ██║██╔████╔██║██████╦╝██████╔╝██║██║░░░██║██╔████╔██║
-        ██║██║╚██╔╝██║██╔══██╗██╔══██╗██║██║░░░██║██║╚██╔╝██║
-        ██║██║░╚═╝░██║██████╦╝██║░░██║██║╚██████╔╝██║░╚═╝░██║
-        ╚═╝╚═╝░░░░░╚═╝╚═════╝░╚═╝░░╚═╝╚═╝░╚═════╝░╚═╝░░░░░╚═╝
+                              ██╗███╗░░░███╗██████╗░██████╗░██╗██╗░░░██╗███╗░░░███╗
+                              ██║████╗░████║██╔══██╗██╔══██╗██║██║░░░██║████╗░████║
+                              ██║██╔████╔██║██████╦╝██████╔╝██║██║░░░██║██╔████╔██║
+                              ██║██║╚██╔╝██║██╔══██╗██╔══██╗██║██║░░░██║██║╚██╔╝██║
+                              ██║██║░╚═╝░██║██████╦╝██║░░██║██║╚██████╔╝██║░╚═╝░██║
+                              ╚═╝╚═╝░░░░░╚═╝╚═════╝░╚═╝░░╚═╝╚═╝░╚═════╝░╚═╝░░░░░╚═╝
 
 
         ## Basics
 
         This library aims to ease the application of deep learning models for time
-        series forecasting. To achieve this, the library differentiates between two
+        series forecasting. Multiple architectures are offered with a fixed
+        number of layers however the user has full control over the number of neurons
+        per layer, activation function type, loss function type, optimizer type and
+        metrics applied.
+        The library differentiates between two
         modes:
 
         1. Univariate-Multistep forecasting
@@ -92,18 +96,18 @@ Description: # Imbrium
 
         # Choose between one of the architectures:
 
-        # predictor.create_mlp(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_rnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_lstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_gru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_cnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_birnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_bilstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_bigru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_encdec_rnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_encdec_lstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_encdec_cnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_encdec_gru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
+        # predictor.create_mlp(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (50, 'relu'), 'layer1': (25,'relu'), 'layer2': (25, 'relu')})
+        # predictor.create_rnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (40, 'relu'), 'layer1': (50,'relu'), 'layer2': (50, 'relu')})
+        # predictor.create_lstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (40, 'relu'), 'layer1': (50,'relu'), 'layer2': (50, 'relu')})
+        # predictor.create_gru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (40, 'relu'), 'layer1': (50,'relu'), 'layer2': (50, 'relu')})
+        # predictor.create_cnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu')})
+        # predictor.create_birnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (50, 'relu'), 'layer1': (50, 'relu')})
+        # predictor.create_bilstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (50, 'relu'), 'layer1': (50, 'relu')})
+        # predictor.create_bigru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (50, 'relu'), 'layer1': (50, 'relu')})
+        # predictor.create_encdec_rnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (100, 'relu'), 'layer1': (50, 'relu'), 'layer2': (50, 'relu'), 'layer3': (100, 'relu')})
+        # predictor.create_encdec_lstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (100, 'relu'), 'layer1': (50, 'relu'), 'layer2': (50, 'relu'), 'layer3': (100, 'relu')})
+        # predictor.create_encdec_cnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu'), 'layer4': (100, 'relu')})
+        # predictor.create_encdec_gru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (100, 'relu'), 'layer1': (50, 'relu'), 'layer2': (50, 'relu'), 'layer3': (100, 'relu')})
 
         # Fit the predictor object
         predictor.fit_model(epochs: int, show_progress: int = 1, validation_split: float = 0.20, batch_size: int = 10)
@@ -137,12 +141,12 @@ Description: # Imbrium
 
         # Choose between one of the architectures:
 
-        # predictor.create_cnnrnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_cnnlstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_cnngru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_cnnbirnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_cnnbilstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_cnnbigru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
+        # predictor.create_cnnrnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error',  layer_config = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu'), 'layer4': (25, 'relu')})
+        # predictor.create_cnnlstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu'), 'layer4': (25, 'relu')})
+        # predictor.create_cnngru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu'), 'layer4': (25, 'relu')})
+        # predictor.create_cnnbirnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu'), 'layer4': (25, 'relu')})
+        # predictor.create_cnnbilstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu'), 'layer4': (25, 'relu')})
+        # predictor.create_cnnbigru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu'), 'layer4': (25, 'relu')})
 
         # Fit the predictor object
         predictor.fit_model(epochs: int, show_progress: int = 1, validation_split: float = 0.20, batch_size: int = 10)
@@ -179,18 +183,18 @@ Description: # Imbrium
 
         # Choose between one of the architectures:
 
-        # predictor.create_mlp(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_rnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_lstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_gru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_cnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_birnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_bilstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_bigru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_encdec_rnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_encdec_lstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_encdec_cnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_encdec_gru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
+        # predictor.create_mlp(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (50, 'relu'), 'layer1': (25,'relu'), 'layer2': (25, 'relu')})
+        # predictor.create_rnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (40, 'relu'), 'layer1': (50,'relu'), 'layer2': (50, 'relu')})
+        # predictor.create_lstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (40, 'relu'), 'layer1': (50,'relu'), 'layer2': (50, 'relu')})
+        # predictor.create_gru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (40, 'relu'), 'layer1': (50,'relu'), 'layer2': (50, 'relu')})
+        # predictor.create_cnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu')})
+        # predictor.create_birnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (50, 'relu'), 'layer1': (50, 'relu')})
+        # predictor.create_bilstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (50, 'relu'), 'layer1': (50, 'relu')})
+        # predictor.create_bigru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (50, 'relu'), 'layer1': (50, 'relu')})
+        # predictor.create_encdec_rnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (100, 'relu'), 'layer1': (50, 'relu'), 'layer2': (50, 'relu'), 'layer3': (100, 'relu')})
+        # predictor.create_encdec_lstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (100, 'relu'), 'layer1': (50, 'relu'), 'layer2': (50, 'relu'), 'layer3': (100, 'relu')})
+        # predictor.create_encdec_cnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu'), 'layer4': (100, 'relu')})
+        # predictor.create_encdec_gru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config: dict = {'layer0': (100, 'relu'), 'layer1': (50, 'relu'), 'layer2': (50, 'relu'), 'layer3': (100, 'relu')})
 
         # Fit the predictor object
         predictor.fit_model(epochs: int, show_progress: int = 1, validation_split: float = 0.20, batch_size: int = 10)
@@ -224,12 +228,12 @@ Description: # Imbrium
 
         # Choose between one of the architectures:
 
-        # predictor.create_cnnrnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_cnnlstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_cnngru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_cnnbirnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_cnnbilstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
-        # predictor.create_cnnbigru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error')
+        # predictor.create_cnnrnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu'), 'layer4': (25, 'relu')})
+        # predictor.create_cnnlstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu'), 'layer4': (25, 'relu')})
+        # predictor.create_cnngru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu'), 'layer4': (25, 'relu')})
+        # predictor.create_cnnbirnn(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu'), 'layer4': (25, 'relu')})
+        # predictor.create_cnnbilstm(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error',  layer_config = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu'), 'layer4': (25, 'relu')})
+        # predictor.create_cnnbigru(optimizer: str = 'adam', loss: str = 'mean_squared_error', metrics: str = 'mean_squared_error', layer_config = {'layer0': (64, 1, 'relu'), 'layer1': (32, 1, 'relu'), 'layer2': (2), 'layer3': (50, 'relu'), 'layer4': (25, 'relu')})
 
         # Fit the predictor object
         predictor.fit_model(epochs: int, show_progress: int = 1, validation_split: float = 0.20, batch_size: int = 10)