Anomaly Detection

0.0(0)
studied byStudied by 0 people
learnLearn
examPractice Test
spaced repetitionSpaced Repetition
heart puzzleMatch
flashcardsFlashcards
Card Sorting

1/6

flashcard set

Earn XP

Description and Tags

finding outlier

Study Analytics
Name
Mastery
Learn
Test
Matching
Spaced

No study sessions yet.

7 Terms

1
New cards

Anomaly Detection:

to detect the outlier`

we can detect this using the isolation tree 🎄 🎄

<p>to detect the outlier`</p><p></p><p>we can detect this using the isolation tree <span data-name="christmas_tree" data-type="emoji">🎄</span> <span data-name="christmas_tree" data-type="emoji">🎄</span> </p>
2
New cards

Local Outlier vs Global Outlier

This is near some Cluster

Not Near to cluster

<p>This is near some Cluster</p><p></p><p>Not Near to cluster</p>
3
New cards

Isolation of tree with 1 outlier

knowt flashcard image
4
New cards

Isolation of tree with 2 outlier

knowt flashcard image
5
New cards

if contamination float value > _____ consider as outlier

0.5

6
New cards

we can also find outler using _______ Algo

DBSCAN

plt.scatter(X[:,0],X[:,1],c=dbcan.labels_)
# dark violet is the outlier

<p>DBSCAN<br></p><pre><code class="language-python">plt.scatter(X[:,0],X[:,1],c=dbcan.labels_) # dark violet is the outlier</code></pre><p></p>
7
New cards

How do we use local outlier factor 

plt.scatter(X[y_pred == -1, 0], X[y_pred == -1, 1], color='r', s=6., label='Detected anomalies')
  # s=6.: Sets the size of the points to 6.