Fungsi loss Bagian I

Merancang Alur Kerja Machine Learning di Python

Dr. Chris Anagnostopoulos

Honorary Associate Professor

Dataset KDD '99 cup

kdd.iloc[0]

kdd.iloc[0]
duration                         51
protocol_type                   tcp
service                        smtp
flag                             SF
src_bytes                      1169
dst_bytes                       332
land                              0
...
dst_host_rerror_rate              0
dst_host_srv_rerror_rate          0
label                          good

False positive vs false negative

Binerkan label:

kdd['label'] = kdd['label'] == 'bad'

Latih klasifier Gaussian Naive Bayes:

clf = GaussianNB().fit(X_train, y_train)
predictions = clf.predict(X_test)
results = pd.DataFrame({
    'actual': y_test,
    'predicted': predictions
})

False positive vs false negative

Binerkan label:

kdd['label'] = kdd['label'] == 'bad'

Latih klasifier Gaussian Naive Bayes:

clf = GaussianNB().fit(X_train, y_train)
predictions = clf.predict(X_test)
results = pd.DataFrame({
    'actual': y_test,
    'predicted': predictions
})

Ada empat kemungkinan konfigurasi label vs prediksi: keduanya True, keduanya False, label True dengan prediksi False, dan label False dengan prediksi True. Kombinasi terakhir disorot di sini.

False positive vs false negative

Binerkan label:

kdd['label'] = kdd['label'] == 'bad'

Latih klasifier Gaussian Naive Bayes:

clf = GaussianNB().fit(X_train, y_train)
predictions = clf.predict(X_test)
results = pd.DataFrame({
    'actual': y_test,
    'predicted': predictions
})

Sekarang, kombinasi label True dan prediksi False yang disorot.

False positive vs false negative

Binerkan label:

kdd['label'] = kdd['label'] == 'bad'

Latih klasifier Gaussian Naive Bayes:

clf = GaussianNB().fit(X_train, y_train)
predictions = clf.predict(X_test)
results = pd.DataFrame({
    'actual': y_test,
    'predicted': predictions
})

Dua kasus saat prediksi sesuai dengan label kini disorot.

Confusion matrix

conf_mat = confusion_matrix(
    ground_truth, predictions)

array([[9477,   19],
       [ 397, 2458]])

tn, fp, fn, tp = conf_mat.ravel()
(fp, fn)

(19, 397)

Sebuah confusion matrix yang menghitung kasus untuk masing-masing dari empat kombinasi yang disebutkan sebelumnya pada dataset ini.

Metrik kinerja skalar

accuracy = 1-(fp + fn)/len(ground_truth)

recall = tp/(tp+fn)

fpr = fp/(tn+fp)

precision = tp/(tp+fp)

f1 = 2*(precision*recall)/(precision+recall)

accuracy_score(ground_truth, predictions)
recall_score(ground_truth, predictions)
precision_score(ground_truth, predictions)
f1_score(ground_truth, predictions)

False positive vs false negative

Klasifier A:

tn, fp, fn, tp = confusion_matrix(
    ground_truth, predictions_A).ravel()
(fp,fn)

(3, 3)

cost = 10 * fp + fn

Klasifier B:

tn, fp, fn, tp = confusion_matrix(
    ground_truth, predictions_B).ravel()
(fp,fn)

(0, 26)

cost = 10 * fp + fn

Klasifier mana yang lebih baik?

Merancang Alur Kerja Machine Learning di Python