feat: Second version of VizSet

visualization.py

@@ -1,13 +1,26 @@
+##################
+# import library #
+##################
+
+import pandas as pd
 import numpy as np
 import matplotlib.pyplot as plt
 import seaborn as sns
 import os
 
 from sklearn.metrics import confusion_matrix
-from sklearn.metrics import classification_report
+from sklearn.metrics import accuracy_score, f1_score, recall_score, auc
+
+#################
+# set variables #
+#################
 
 VIZ_SAVE_PATH = "./viz_results"
 
+#####################
+# class & functions #
+#####################
+
 class VizSet:
     def __init__(self, real, pred):
         self.real = real
@@ -28,14 +41,14 @@ def plot_confusion_matrix(
             **key_args
         )->None:
         """
-            Plot or save confusion matrix
+            Plot or save confusion matrix image
+
             Parameters:
                 real : np.array, ground trues of label if None 
                 pred : np.array, model predicted results
                 save : bool, defalut = True, save or plot options : {save, linux, window, notebook}
-                file_name : str, save image name containg extentions
-
-                other parameter is available for the same function, "seaborn heatmap"
+                file_name : str, save image name containg extentions and path, default = "./viz_results/cm.jpg"
+                title : str, save image's title ex)"Epoch 1 result CM"
         """
         # check input data
         if len(real) < 1:
@@ -73,13 +86,21 @@ def plot_confusion_matrix(
             cmap='Blues',
             fmt='d',
             cbar=False,
-            **key_args,
         )
 
         # set axis annotation
-        plt.title("Confusion Matrix")
-        plt.xlabel("Real")
-        plt.ylabel("Pred")
+        args_dict = {
+            "title" : "Confusion Matrix",
+            "xlabel" : "Real",
+            "ylabel" : "Pred"
+        }
+        for key in args_dict.keys():
+            if key in key_args:
+                args_dict[key] = key_args[key]
+
+        plt.title(args_dict['title'])
+        plt.xlabel(args_dict['xlabel'])
+        plt.ylabel(args_dict['ylabel'])
 
         # save or plot result
         if save:
@@ -88,3 +109,148 @@ def plot_confusion_matrix(
         else:
             plt.show()
 
+
+    def plot_metrics(
+        self,
+        real : np.array = [],
+        pred : np.array = [],
+        save = True,
+        title = "Metrics",
+        file_name = "metrics.jpg",
+        **metrics,
+    )->None:
+        """
+            Plot or save selected metrics' bar chart image
+
+            Parameters:
+                Base values : Array, Essential values for calculate metrics
+                ------------------------------------------------------
+                    real : np.array, ground trues value list
+                    pred : np.array, model predicted value list
+                    
+                key values : bool, Optional plotting to chart
+                ------------------------------------------------------
+                    Available options:
+                        accuracy, f1, precision, recall, EM, AUC, BLEU
+                
+                accuracy  : bool, calculate accuracy
+                f1        : bool, calculate f1-micro score
+                precision : bool, calculate precision
+                recall    : bool, calculate recall
+                EM        : bool, calculate Exact Match
+                AUC       : bool, calculate auc score
+
+            Example:
+            >>> vizset = VizSet(real, pred)
+            >>> vizset.plot_metrics(accuracy=True, EM=True, f1=True)
+        """
+        # check input data
+        if len(real) < 1:
+            real = self.real
+        if len(pred) < 1:
+            pred = self.pred
+
+        # set calculations
+        cal_metrics = {
+            "accuracy" : accuracy_score,
+            "f1" :       f1_score,
+            "recall" :   recall_score,
+            "AUC" :      auc,
+            "EM" :       EM_score,
+        }
+
+        # calculate metrics
+        result = {}
+        for key in metrics.keys():
+            if key == "f1":
+                args = {"average":"micro"}
+                result[key] = cal_metrics[key](real, pred, **args)
+                continue
+
+            result[key] = cal_metrics[key](real, pred)
+
+
+        # plot metrics
+        width, height = 4, 6
+
+        fig = plt.figure(
+            figsize=(width, height)
+        )
+        plt.title(title)
+        metric_df = pd.DataFrame({key:[val] for key,val in result.items()})
+        sns.barplot(data=metric_df)
+
+        # save or plot result
+        if save:
+            save_path = os.path.join(VIZ_SAVE_PATH, file_name)
+            fig.savefig(save_path)
+        else:
+            plt.show()
+
+        return result
+
+def EM_score(
+    real:np.array, 
+    pred:np.array
+)->float:
+    counter = 0
+    for r, p in zip(real, pred):
+        if r == p:
+            counter+=1
+    return counter / len(real)
+
+
+def print_confusion_matrix(
+    real:np.array, 
+    pred:np.array, 
+    label:dict=None
+)->None:
+    """
+        Print confusion matrix for console
+
+        Parameters:
+            real : np.array, real value list
+            pred : np.array, predicted value list by model
+            label : dict, default = None, print which idx is label
+
+        Example:
+        >>> print_confusion_matrix(np.array([1,2,3,4]),np.array([1,2,3,4]))
+        0|  1  0  0  0
+        1|  0  1  0  0
+        2|  0  0  1  0
+        3|  0  0  0  1
+          ------------
+            0  1  2  3
+        
+    """
+    # print label's state
+    if label:
+        for key, val in label.items():
+            print(f"{val} : {key}")
+
+    # calculate confusionmatrix
+    cm = confusion_matrix(real, pred)
+
+    # print title
+    print("confusion matrix")
+
+    #print grid
+    print("   "+"-"*5*cm.shape[0])
+    for idx in range(cm.shape[0]):
+        print("%2d|"%idx,end="")
+        for jdx in range(cm.shape[1]):
+            print("%5d" % cm[idx, jdx], end="")
+        print()
+
+    print("   ",end="")
+    print("-"*5*cm.shape[0], end="")
+    print()
+    print("   ",end="")
+    for idx in range(cm.shape[0]):
+        print("%5d"%idx, end="")
+    print()
+
+
+
+
+