.

v1.py

@@ -144,7 +144,7 @@ model = keras.Sequential(
         layers.Input(shape=(x_train[0].shape[1], x_train[0].shape[2])),
         layers.Conv1D(
             filters=NumFilters,
-            kernel_size=7,
+            kernel_size=KernelSize,
             padding="same",
             strides=2,
             activation="relu",

v1_multifailure.py

@@ -123,7 +123,7 @@ def plotData():
 NumFilters=64
 KernelSize=7
 DropOut=0.2
-ThresholdFactor=1.7
+ThresholdFactor=2
 TIME_STEPS = 48
 def create_sequences(values, time_steps=TIME_STEPS):
     output = []
@@ -239,6 +239,7 @@ for i in range(1,NumberOfFailures+1):
     rnext=r+dataTestNorm[i].shape[0]
     testRanges.append([r,rnext] )
     r=rnext
+
 testRanges.append([r, x_test.shape[0]+TIME_STEPS  ])
 
 
@@ -345,7 +346,6 @@ def anomalyMetric(testList):  # first of list is non failure data
 anomalyMetric([dataTestNorm[0],dataTestNorm[1],dataTestNorm[2],dataTestNorm[3],dataTestNorm[4]])
 plotData2()
 
-exit(0)
 #   2nd scenario. Go over anomalies and classify it by less error
 '''   
 #This code works, but too slow
@@ -370,17 +370,20 @@ for i in anomalous_data_indices:
     anomalous_data_type.append(np.argmin(error)+1)
 
 
-# For plotting purposes
-
-
 anomalous_data_indices_by_failure=[]
 for i in range(NumberOfFailures+1):
     anomalous_data_indices_by_failure.append([])
 
 for i in range(len(anomalous_data_indices)):
-    print(i," ",anomalous_data_type[i])
+    #print(i," ",anomalous_data_type[i])
     anomalous_data_indices_by_failure[anomalous_data_type[i]].append(anomalous_data_indices[i])  
 
+MaxIndex=0
+for i in range(1,NumberOfFailures+1):
+    MaxIndex=max(MaxIndex,max(anomalous_data_indices_by_failure[i]))
+
+# Enlarge x_test to plot failure points
+XTest=np.vstack((x_test[:,0,:],x_test[-1,1:TIME_STEPS,:]))
 
 def plotData3():
     NumFeaturesToPlot=len(indexesToPlot)
@@ -391,22 +394,21 @@ def plotData3():
         init=0
         end=len(x_train[0])
         axes[i].plot(range(init,end),x_train[0][:,0,indexesToPlot[i]],label="normal train")
-        #axes.plot(range(len(x_train[0]),len(x_train[0])+len(x_test)),x_test[:,0,0],label="abnormal")
         init=end
         end+=testRanges[0][1]
-        axes[i].plot(range(init,end),x_test[testRanges[0][0]:testRanges[0][1],0,indexesToPlot[i]],label="normal test")
+        axes[i].plot(range(init,end),XTest[testRanges[0][0]:testRanges[0][1],indexesToPlot[i]],label="normal test")
         init=end
         end+=(testRanges[1][1]-testRanges[1][0])
         for j in range(1,NumberOfFailures+1):
-            axes[i].plot(range(init,end),x_test[testRanges[j][0]:testRanges[j][1],0,indexesToPlot[i]],label="fail type "+str(j), color=colorline[j-1])
+            axes[i].plot(range(init,end),XTest[testRanges[j][0]:testRanges[j][1],indexesToPlot[i]],label="fail type "+str(j), color=colorline[j-1])
             init=end
             end+=(testRanges[j+1][1]-testRanges[j+1][0])
-
-            axes[i].plot(len(x_train[0])+np.array(anomalous_data_indices_by_failure[j])+TIME_STEPS,x_test[np.array(anomalous_data_indices_by_failure[j])+TIME_STEPS,0,indexesToPlot[i]],color=colordot[j-1],marker='.',linewidth=0,label="abnormal detection type "+str(j))
+            ##   MODIFY here as in PRevious Plot
+            axes[i].plot(len(x_train[0])+np.array(anomalous_data_indices_by_failure[j])+TIME_STEPS,XTest[np.array(anomalous_data_indices_by_failure[j])+TIME_STEPS,indexesToPlot[i]],color=colordot[j-1],marker='.',linewidth=0,label="abnormal detection type "+str(j))
 
         init=end-(testRanges[NumberOfFailures+1][1]-testRanges[NumberOfFailures+1][0])
         end=init+(testRanges[0][1]-testRanges[0][0])
-        axes[i].plot(range(init,end),x_test[testRanges[0][0]:testRanges[0][1],0,indexesToPlot[i]],color='orange')
+        axes[i].plot(range(init,end),XTest[testRanges[0][0]:testRanges[0][1],indexesToPlot[i]],color='orange')
 
         if i==0:
             axes[i].legend(bbox_to_anchor=(1, 0.5))
@@ -418,3 +420,8 @@ def plotData3():
 
 anomalyMetric([dataTestNorm[0],dataTestNorm[1],dataTestNorm[2],dataTestNorm[3],dataTestNorm[4]])
 plotData3()
+
+#  A new anomalyMEtric for multiclass must be defined
+    # look at anomalies detected on first stage
+    # l0ok at classification of second stage 
+    # then: determine if: correct classified  or  bad classified/unclassified