accuracy for application

Author: mufwan

scripts/keywordSpotting.dml

@@ -1,5 +1,4 @@
 
-
 X = read("tmp/waves.csv", format="csv")
 y = read("tmp/labels.csv", format="csv")
 
@@ -11,42 +10,25 @@ Xy_shuffled = Xy_shuffled[,1:num_col] # Remove the random column
 X_shuffled = Xy_shuffled[,1:(num_col-1)] # Features, excluding the last column which is labels
 y_shuffled = Xy_shuffled[,num_col] # Labels, the last column
 
-
 [X_train, X_test, y_train, y_test] = split(X=X_shuffled, Y=y_shuffled, seed= 13)
 
-# Preprocess with fft Optionally move before split.
 [X_train_re, X_train_im] = stft(X_train, 4, 2)
 [X_test_re, X_test_im] = stft(X_test, 4, 2)
 
-
 X_train_re_sq = X_train_re^2
 X_train_im_sq = X_train_im^2
 
-
 X_test_re_sq = X_test_re^2
 X_test_im_sq = X_test_im^2
 
-
-# Sum the squared matrices
 sum_X_train_sq = X_train_re_sq + X_train_im_sq
 sum_X_test_sq = X_test_re_sq + X_test_im_sq
 
-# Compute the square root of each element in the sum matrix to get the magnitudes
 magnitudes_train = sqrt(sum_X_train_sq)
 magnitudes_test = sqrt(sum_X_test_sq)
 
 bias = lm(X=magnitudes_train, y=y_train, reg=1e-1, maxi=10, verbose=TRUE)
 predictions = lmPredict(X=magnitudes_test, B=bias, verbose=FALSE)
 
-#sign_predictions = sign(predictions)
-#sign_Y = sign(y_test)
-
-#correct = sum(sign_predictions == sign_Y)
-#total = nrow(y_test)
-#accuracy = correct / total
-
-#print(toString(predictions))
-#print(toString(accuracy))
-
-R = lmPredictStats(yhat=predictions, ytest=y_test, lm=TRUE)
-print(toString(R))
+accuracy = auc(Y=y_test, P=predictions)
+print(toString(accuracy))