Hwallace/jupyter (#35)

* updates to include a notebook interface

* Remove debug path

* Add sklearn notebook

* more notebook tuning

* update basic

* add pytorch

* Delete notebooks/.pdf.png

Author: henry-wallace-phys

notebooks/basic_network.ipynb

@@ -0,0 +1,78 @@
+from MaCh3PythonUtils.file_handling.chain_handler import ChainHandler
+import matplotlib.pyplot as plt
+
+class ChainDiagnostics:
+    def __init__(self, config_reader: ChainHandler) -> None:
+        self._chain_handler = config_reader
+        
+    def _extract_chain_information(self, parameter_name: str | int):
+        if isinstance(parameter_name, str):
+            parameter_id = self._chain_handler.ttree_array.columns.get_loc(parameter_name)
+        if isinstance(parameter_name, int):
+            parameter_id = parameter_name
+            parameter_name = self._chain_handler.ttree_array.columns[parameter_id]
+        
+        return self._chain_handler.ttree_array.iloc[:,parameter_id], parameter_name
+        
+    def __make_plot(self, fig, axs):
+        if fig is None:
+            fig, axs = plt.subplots(1, 1, figsize=(10, 5))
+        elif axs is None:
+            axs = fig.add_subplot(1, 1, 1)
+        
+        return fig, axs
+        
+    def make_trace_plot(self, parameter_name: str | int, axs=None, fig =None):
+        fig, axs = self.__make_plot(fig, axs)
+
+        chain, parameter_name = self._extract_chain_information(parameter_name)
+        axs.plot(chain, linewidth=0.5, color='darkorange')
+
+        return fig, axs
+    
+    def make_autocorr_plot(self, parameter_name: str | int, axs=None, fig =None):
+        fig, axs = self.__make_plot(fig, axs)
+
+        if fig is None:
+            fig, axs = plt.subplots(1, 1, figsize=(10, 5))
+        elif axs is None:
+            axs = fig.add_subplot(1, 1, 1)
+
+        chain, parameter_name = self._extract_chain_information(parameter_name)
+        axs.acorr(chain, maxlags=1000, linewidth=0.5, color='darkorange')
+
+        return fig, axs
+    
+    def make_posterior_hist_plot(self, parameter_name: str | int, axs=None, fig =None, is_horizontal=False):
+        fig, axs = self.__make_plot(fig, axs)
+
+        if fig is None:
+            fig, axs = plt.subplots(1, 1, figsize=(10, 5))
+        elif axs is None:
+            axs = fig.add_subplot(1, 1, 1)
+        
+        orientation = 'vertical'
+
+        if is_horizontal:
+            orientation = 'horizontal'
+        
+        chain, parameter_name = self._extract_chain_information(parameter_name)    
+        axs.hist(chain, bins=50, density=True, linewidth=0.5, color='darkorange', alpha=0.5, orientation=orientation)
+
+        return fig, axs
+    
+    def __call__(self, parameter_name: str):
+        fig, axs = plt.subplots(2, 2, figsize=(15, 5))
+        axs[1][1].remove()
+        axs[1][0].remove()
+
+        fig, axs[0][0] = self.make_trace_plot(parameter_name, axs=axs[0][0], fig=fig)
+        fig, axs[0][1] = self.make_posterior_hist_plot(parameter_name, axs=axs[0][1], fig=fig, is_horizontal=True)
+
+        # To share the same axis etc,
+        plt.setp(axs[0][1].get_yticklabels(), visible=False)
+        fig.subplots_adjust(wspace=.0)
+
+
+        # fig, axs[1][0] = self.make_autocorr_plot(parameter_name, axs=axs[1][0], fig=fig)
+        return fig, axs

notebooks/basic_random_forest.ipynb

@@ -20,6 +20,8 @@
 import matplotlib.pyplot as plt
 from matplotlib.backends.backend_pdf import PdfPages
 
+from rich import print
+
 class FileMLInterface(ABC):
     white_viridis = LinearSegmentedColormap.from_list('white_viridis', [
         (0, '#ffffff'),
@@ -59,7 +61,7 @@ def __init__(self, chain: ChainHandler, prediction_variable: str, fit_name: str)
         self._scaler = StandardScaler()
         # self._pca_matrix = PCA(n_components=0.95)
 
-        self._label_scaler = MinMaxScaler(feature_range=(0, 1))
+        self._label_scaler = StandardScaler()
 
 
 
@@ -86,19 +88,19 @@ def set_training_test_set(self, test_size: float):
         self._training_data, self._test_data, self._training_labels, self._test_labels =  train_test_split(features, labels, test_size=test_size)
 
         # Fit scaling pre-processors. These get applied properly when scale_data is called
-        _= self._scaler.fit_transform(self._training_data)
-        self._label_scaler.fit_transform(self._training_labels)
+        self._scaler.fit(self._training_data)
+        self._label_scaler.fit(self._training_labels)
 
         # self._pca_matrix.fit(scaled_training)
 
     def scale_data(self, input_data):
         # Applies transformations to data set
         scale_data = self._scaler.transform(input_data)
-        # scale_data = self._pca_matrix.transform(scale_data)
         return scale_data
 
     def scale_labels(self, labels):
         return self._label_scaler.transform(labels)
+        # return labels.values.reshape(-1, 1)
 
     def invert_scaling(self, input_data):
         # Inverts transform
@@ -193,7 +195,7 @@ def load_model(self, input_model: str):
         :param input_file: Pickled Model
         :type input_file: str
         """        
-        print(f"Attempting to load file from {input_file}")
+        print(f"[spring_green1]Attempting to load file from[/spring_green1][bold red3] {input_file}")
         with open(input_model, 'r') as f:
             self._model = pickle.load(f)
 
@@ -216,14 +218,18 @@ def test_model(self):
         train_as_numpy = self.scale_labels(self._training_labels).T[0]
         self.evaluate_model(train_prediction, train_as_numpy, "train_qq_plot.pdf")
 
-        print("=====\n\n")
+        print("=====")
         print("Testing Results!")
 
         test_prediction = self.model_predict(self._test_data)
         test_as_numpy = self.scale_labels(self._test_labels).T[0]
 
         self.evaluate_model(test_prediction, test_as_numpy, outfile=f"{self._fit_name}")
-        print("=====\n\n")
+        print("=====")
+        
+
+    def print_model_summary(self):
+        print("Model Summary")
 
     def model_predict_single_sample(self, sample):
         sample_shaped = sample.reshape(1,-1)
@@ -232,7 +238,7 @@ def model_predict_single_sample(self, sample):
     def get_maxlikelihood(self)->OptimizeResult:
         init_vals = self.training_data.iloc[[1]].to_numpy()[0]
 
-        print("Calculating max LLH")
+        print("[bold purple]Calculating max LLH")
         maximal_likelihood = minimize(self.model_predict_single_sample, init_vals, bounds=zip(self._chain.lower_bounds[:-1], self._chain.upper_bounds[:-1]), method="L-BFGS-B", options={"disp": True})
         return maximal_likelihood
 
@@ -245,9 +251,9 @@ def run_likelihood_scan(self, n_divisions: int = 500):
 
         errors = np.sqrt(np.diag(maximal_likelihood.hess_inv(np.identity(self.chain.ndim-1))))
 
-        print("Maximal Pars :")
+        print("[bold red3]Maximal Pars :")
         for i in range(self.chain.ndim-1):
-            print(f"Param : {self.chain.plot_branches[i]} : {maximal_likelihood.x[i]}±{errors[i]}")
+            print(f"[bold red3]Param :[/bold red3] [yellow3]{self.chain.plot_branches[i]} : {maximal_likelihood.x[i]}±{errors[i]}")
 
 
         with PdfPages("llh_scan.pdf") as pdf:
@@ -285,13 +291,14 @@ def evaluate_model(self, predicted_values: Iterable, true_values: Iterable, outf
         :type outfile: str, optional
         """                
 
-        print(predicted_values)
-        print(f"Mean Absolute Error : {metrics.mean_absolute_error(predicted_values,true_values)}")
-        
+        print(f"[bold red3]Mean Absolute Error :[/bold red3] [yellow3]{metrics.mean_absolute_error(predicted_values,true_values)}")
 
+        outfile_name = outfile.split(".")[0]
+        outfile = f"{outfile_name}.pdf"
+        warnings.filterwarnings("ignore", message="Polyfit may be poorly conditioned")
         lobf = np.poly1d(np.polyfit(predicted_values, true_values, 1))
 
-        print(f"Line of best fit : y={lobf.c[0]}x + {lobf.c[1]}")
+        print(f"[bold purple]Line of best fit :[/bold purple] [dodger_blue1]y={lobf.c[0]}x + {lobf.c[1]}")
 
         fig = plt.figure()
 
@@ -322,11 +329,21 @@ def evaluate_model(self, predicted_values: Iterable, true_values: Iterable, outf
         ax.set_ylabel("True Log Likelihood")
 
         fig.legend()
+        
         if outfile=="": outfile = f"evaluated_model_qq_tf.pdf"
 
-        print(f"Saving QQ to {outfile}")
+        print(f"[bold spring_green1]Saving QQ to[/bold spring_green1][dodger_blue1] {outfile}")
 
         fig.savefig(outfile)
+        
+        try:
+            is_notebook = self.is_notebook()
+            if is_notebook:
+                plt.show()
+        except Exception:
+            ...
+            
+        
         plt.close()
 
         # Gonna draw a hist
@@ -335,4 +352,18 @@ def evaluate_model(self, predicted_values: Iterable, true_values: Iterable, outf
         plt.hist(difs, bins=100, density=True, range=(np.std(difs)*-5, np.std(difs)*5))
         plt.xlabel("True - Pred")
         plt.savefig(f"diffs_5sigma_range_{outfile}")
-        plt.close()
+        
+        plt.close()
+        
+    @classmethod
+    def is_notebook(cls) -> bool:
+        try:
+            shell = get_ipython().__class__.__name__
+            if shell == 'ZMQInteractiveShell':
+                return True   # Jupyter notebook or qtconsole
+            elif shell == 'TerminalInteractiveShell':
+                return False  # Terminal running IPython
+            else:
+                return False  # Other type (?)
+        except NameError:
+            return False      # Probably standard Python interpreter

notebooks/basic_sklearn.ipynb

@@ -11,8 +11,9 @@
 from MaCh3PythonUtils.machine_learning.tensorflow.tf_manual_interface import TfManualLayeredInterface
 from MaCh3PythonUtils.machine_learning.tensorflow.tf_interface import TfInterface
 
-from MaCh3PythonUtils.file_handling.chain_handler import ChainHandler
+from MaCh3PythonUtils.machine_learning.torch.torch_interface import TorchInterface
 
+from MaCh3PythonUtils.file_handling.chain_handler import ChainHandler
 import sklearn.ensemble as ske
 import tensorflow.keras as tfk
 
@@ -32,6 +33,9 @@ class MLFactory:
             "normalizing_flow": TfNormalizingFlowModel,
             "autotune": TfAutotuneInterface
         },
+        "torch": {
+            "sequential": TorchInterface
+        }
     }
 
     def __init__(self, input_chain: ChainHandler, prediction_variable: str, plot_name: str):
@@ -92,7 +96,7 @@ def __make_scikit_model(self, algorithm: str, **kwargs)->SciKitInterface:
     def __make_tensorflow_layered_model(self, interface: TfManualLayeredInterface, layers: dict)->TfManualLayeredInterface:
         for layer in layers:
             layer_id = list(layer.keys())[0]                
-            interface.add_layer(layer_id, layer[layer_id])
+            interface.add_layer(layer_id, layer[layer_id].copy())
 
         return interface
 
@@ -106,14 +110,28 @@ def __make_tensorflow_model(self, algorithm: str, **kwargs)->TfInterface:
 
         # Ugh
         if algorithm=="sequential" or algorithm=="residual":
-            print("HERE")
             model = self.__make_tensorflow_layered_model(model, kwargs["Layers"])
             model.set_training_settings(kwargs.get("FitSettings"))
 
 
         model.build_model(**kwargs["BuildSettings"])
 
         return model
+    
+    def __make_torch_model(self, algorithm: str, **kwargs)->TorchInterface:
+        model_func = self.__IMPLEMENTED_ALGORITHMS["torch"].get(algorithm.lower(), None)
+        
+        if model_func is None:
+            raise Exception(f"Cannot find {algorithm}")
+        
+        model: TorchInterface = model_func(self._chain, self._prediction_variable, self._plot_name)
+
+        for layer in kwargs["Layers"]:
+            layer_id = list(layer.keys())[0]
+            model.add_layer(layer_id=layer_id, layer_args=layer[layer_id].copy())
+        
+        model.build_model(**kwargs["BuildSettings"], **kwargs["FitSettings"])
+        return model
 
     def make_interface(self, interface_type: str, algorithm: str, **kwargs):
         interface_type = interface_type.lower()
@@ -122,5 +140,9 @@ def make_interface(self, interface_type: str, algorithm: str, **kwargs):
                 return self.__make_scikit_model(algorithm, **kwargs)
             case "tensorflow":
                 return self.__make_tensorflow_model(algorithm, **kwargs)
+            case "torch":
+                return self.__make_torch_model(algorithm, **kwargs)
+        
             case _:
-                raise Exception(f"{interface_type} not implemented!")
+                raise Exception(f"{interface_type} not implemented!")
+            

notebooks/basic_torch.ipynb

@@ -1,5 +1,6 @@
 from pandas import DataFrame
 from MaCh3PythonUtils.machine_learning.file_ml_interface import FileMLInterface
+from tqdm import tqdm
 
 """
 TODO: 

src/MaCh3PythonUtils/file_handling/chain_diagnostics.py

@@ -103,4 +103,5 @@ def evaluate_model(self, predicted_values: Iterable, true_values: Iterable, outf
 
         # CODE TO DO TF SPECIFIC PLOTS GOES HERE
 
-        return super().evaluate_model(predicted_values, true_values, outfile)
+        return super().evaluate_model(predicted_values, true_values, outfile)
+    

src/MaCh3PythonUtils/machine_learning/file_ml_interface.py

@@ -22,9 +22,8 @@ def train_model(self):
         scaled_data = self.scale_data(self._training_data)
         scaled_labels = self.scale_labels(self._training_labels)
 
-        lr_schedule = tfk.callbacks.ReduceLROnPlateau(monitor="val_loss", patience=10, factor=0.5, min_lr=1e-8, verbose=1)
+        lr_schedule = tfk.callbacks.ReduceLROnPlateau(monitor="val_loss", patience=10, factor=0.1, min_lr=1e-9, verbose=1)
         stop_early = tfk.callbacks.EarlyStopping(monitor='val_loss', patience=20)
-
         self._model.fit(scaled_data, scaled_labels, **self._training_settings, callbacks=[lr_schedule, stop_early])
 
 
@@ -53,12 +52,6 @@ def add_layer(self, layer_id: str, layer_args: dict):
             # Hacky, swaps string value of regularliser for proper one
             layer_args["kernel_regularizer"] = tfk.regularizers.L2(layer_args["kernel_regularizer"])
 
+
         self._layers.append(self.__TF_LAYER_IMPLEMENTATIONS[layer_id.lower()](**layer_args))
 
-
-
-
-
-
-        
-