Create app.py

app.py

@@ -0,0 +1,152 @@
+
+import numpy as np
+
+import gradio as gr
+
+
+set_input = gr.inputs.Dataframe(type="numpy", row_count=10, col_count=3, headers=['Sample Index', 'Predicted Prob', 'Label (Y)'], datatype=["number", "number", "number"])
+set_input2 = gr.inputs.Slider(0, 1, step = 0.1, default=0.4)
+
+#set_output = gr.inputs.Textbox(label ='test')
+set_output1 = gr.outputs.Dataframe(type="pandas", label = 'Predicted Labels',max_rows=10)
+
+set_output2 = gr.outputs.Image(label="Confusion Matrix")
+set_output3 = gr.outputs.Image(label="ROC curve")
+
+def visualize_ROC(set_threshold,set_input):
+  prob = set_input[:,1]
+  pred_label = (prob >= set_threshold).astype(int)
+  actual_label = set_input[:,2]
+  import pandas as pd
+
+  data = {
+          'Predicted Prob': prob,
+          'Predicted Label': pred_label,
+          'Actual Label':   actual_label
+          }
+
+  import pandas as pd
+  import seaborn as sn
+  import matplotlib.pyplot as plt
+
+
+
+  df = pd.DataFrame(data)
+  confusion_matrix_results = confusion_matrix(df['Actual Label'], df['Predicted Label'])
+
+  fig, ax = plt.subplots(figsize=(12,4))
+  sn.heatmap(confusion_matrix_results, annot=True,annot_kws={"size": 20},cbar=False,
+                  square=False,
+                  fmt='g',
+                  cmap=ListedColormap(['white']), linecolor='black',
+                  linewidths=1.5)
+  
+  sn.set(font_scale=2)
+  plt.xlabel("Predicted Label")
+  plt.ylabel("Actual Label")
+  plt.text(0.6,0.55,'(TN)')
+  plt.text(1.6,0.55,'(FP)')
+  plt.text(0.6,1.55,'(FN)')
+  plt.text(1.6,1.55,'(TP)')
+
+  ax.xaxis.tick_top()
+
+  ax.xaxis.set_ticks_position('top')
+  ax.xaxis.set_label_position('top')
+  plt.tight_layout()
+
+  plt.savefig('tmp.png', dpi=100)
+
+  ## get ROC curve 
+  from sklearn.metrics import roc_curve
+  fpr_mod, tpr_mod, thrsholds_mod = roc_curve(df['Actual Label'], df['Predicted Prob'])
+
+  TP, FP, TN, FN = perf_measure(df['Actual Label'], df['Predicted Label'])
+
+  # Sensitivity, hit rate, recall, or true positive rate
+  try:
+    TPR = TP/(TP+FN)
+  except:
+    TPR = 0
+
+  # Fall out or false positive rate
+  try:
+    FPR = FP/(FP+TN)
+  except:
+    FPR = 0
+
+
+
+  fig, ax = plt.subplots(figsize=(12,8))
+
+  import matplotlib.pyplot as plt
+  import numpy as np
+  plt.rcParams["figure.autolayout"] = True
+  plt.rcParams['figure.facecolor'] = 'white'
+  m1, c1 = 1, 0
+  x = np.linspace(0, 1, 500)
+
+  plt.plot(fpr_mod, tpr_mod, label = 'ROC', c='black', linestyle='-')
+
+  plt.plot(x, x * m1 + c1, 'black', linestyle='--')
+  plt.xlim(0, 1)
+  plt.ylim(0, 1)
+  #xi = (c1 - c2) / (m2 - m1)
+  #yi = m1 * xi + c1
+  plt.axvline(x=FPR, color='gray', linestyle='--')
+  plt.axhline(y=TPR, color='gray', linestyle='--')
+  plt.scatter(FPR, TPR, color='red', s=300)
+
+  ax.set_facecolor("white")
+
+  ax.tick_params(axis='x', colors='black')
+  ax.tick_params(axis='y', colors='black')
+  ax.spines['left'].set_color('black')
+  ax.spines['bottom'].set_color('black')
+  ax.spines['top'].set_color('black')
+  ax.spines['right'].set_color('black')
+  plt.xlabel('False Positive Rate (1 - specificity)')
+  plt.ylabel('True Positive Rate (Recall)')
+  plt.text(TPR, TPR, 'TPR:%s, FPR:%s' % (FPR,TPR))
+  plt.title("ROC curve", fontsize=20)
+  plt.tight_layout()
+
+  plt.savefig('tmp2.png', dpi=100)
+
+  #return df,'tmp.png','tmp2.png'
+  return 'tmp.png','tmp2.png'
+
+def get_example():
+
+  import numpy as np
+  import pandas as pd
+  np.random.seed(seed = 3)
+
+  N=100
+  pd_class1 = pd.DataFrame({'Sample Index': [i for i in range(1,int(N/2)+1)],'Predicted Prob': np.random.uniform(0.3,1,int(N/2)), 'Label (Y)': np.repeat(1,int(N/2))})
+  pd_class2 = pd.DataFrame({'Sample Index': [i for i in range(int(N/2)+1,N+1)],'Predicted Prob': np.random.uniform(0,0.7,int(N/2)), 'Label (Y)': np.repeat(0,int(N/2))})
+
+   
+  pd_all = pd.concat([pd_class1, pd_class2]).reset_index(drop=True)
+  pd_all = pd_all.sample(frac=1)
+  return pd_all.to_numpy()
+
+
+### configure Gradio
+interface = gr.Interface(fn=visualize_ROC, 
+                         inputs=[set_input2, set_input], 
+                         outputs=[set_output2,set_output3],
+                         examples_per_page = 2,
+                         examples=[
+                              [0.5,get_example()],
+                              [0.7,get_example()],
+                          ],
+                         title="CSCI4750/5750: ROC curve", 
+                         description= "Click examples below for a quick demo",
+                         theme = 'huggingface',
+                         layout = 'horizontal',
+                         live=True
+                         )
+
+
+interface.launch(debug=True, height=1400, width=1400)