Trjtdtk

Is it better practice to read straight from sheet music rather than memorize it?

What should you do when eye contact makes your subordinate uncomfortable?

A book (mid 2000s) about a kid getting an old laptop that connects him to aliens, the alien government (I think) coming after him?

Count the occurrence of each unique word in the file

Varistor? Purpose and principle

Calculating Wattage for Resistor in High Frequency Application?

By means of an example, show that P(A) + P(B) = 1 does not mean that B is the complement of A.

What is the difference between "Do you interest" and "...interested in" something?

What should you do if you miss a job interview (deliberately)?

Why does Async/Await work properly when the loop is inside the async function and not the other way around?

Closed-form expression for certain product

If an object with more mass experiences a greater gravitational force, why don't more massive objects fall faster?

Does the expansion of the universe explain why the universe doesn't collapse?

Did US corporations pay demonstrators in the German demonstrations against article 13?

Why did the HMS Bounty go back to a time when whales are already rare?

Freedom of speech and where it applies

Does an advisor owe his/her student anything? Will an advisor keep a PhD student only out of pity?

What spells are affected by the size of the caster?

Is a model fitted to data or is data fitted to a model?

Could Intel SGX be dangerous under Linux?

How should I respond when I lied about my education and the company finds out through background check?

Why did the EU agree to delay the Brexit deadline?

Customize circled numbers

Should I stop contributing to retirement accounts?

How to compare the performance of two unsupervised algorithms on same data-set?

Clustering with Replicator Neural NetworkH2o autoencoder anomaly detection for multivariate time series datahow to compare different sets of time series dataUnsupervised Anomaly Detection in ImagesHow would I apply anomaly detection to time series data in LSTM?Anomaly detection on time seriesAnomaly detection in nominal big dataAnomaly Detection: Model Creation & ImplementationVariable Importance in unsupervised anomaly detection algorithms

5

$begingroup$

I want to solve an anomaly detection problem on an unlabeled data-set. The only information about this problem is that the anomalies population is lower than 0.1%. It should be notice that the size of the feature vector for each sample is 40. Is there any clear way to compare the performance of unsupervised algorithms?

unsupervised-learning anomaly-detection unbalanced-classes evaluation

share|improve this question

edited Mar 20 at 11:07

Alireza Zolanvari

asked Mar 20 at 9:52

Alireza ZolanvariAlireza Zolanvari

35716

$endgroup$

• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  How do you measure performance of single model?
  $endgroup$
  – mikalai
  2 days ago
  

  
  
  
  


• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  @mikalai It is exactly what I have asked
  $endgroup$
  – Alireza Zolanvari
  2 days ago
  

  
  
  
  

add a comment | 

5

$begingroup$

I want to solve an anomaly detection problem on an unlabeled data-set. The only information about this problem is that the anomalies population is lower than 0.1%. It should be notice that the size of the feature vector for each sample is 40. Is there any clear way to compare the performance of unsupervised algorithms?

unsupervised-learning anomaly-detection unbalanced-classes evaluation

share|improve this question

edited Mar 20 at 11:07

Alireza Zolanvari

asked Mar 20 at 9:52

Alireza ZolanvariAlireza Zolanvari

35716

$endgroup$

• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  How do you measure performance of single model?
  $endgroup$
  – mikalai
  2 days ago
  

  
  
  
  


• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  @mikalai It is exactly what I have asked
  $endgroup$
  – Alireza Zolanvari
  2 days ago
  

  
  
  
  

add a comment | 

$begingroup$

I want to solve an anomaly detection problem on an unlabeled data-set. The only information about this problem is that the anomalies population is lower than 0.1%. It should be notice that the size of the feature vector for each sample is 40. Is there any clear way to compare the performance of unsupervised algorithms?

unsupervised-learning anomaly-detection unbalanced-classes evaluation

share|improve this question

edited Mar 20 at 11:07

Alireza Zolanvari

asked Mar 20 at 9:52

Alireza ZolanvariAlireza Zolanvari

35716

$endgroup$

share|improve this question

edited Mar 20 at 11:07

Alireza Zolanvari

asked Mar 20 at 9:52

Alireza ZolanvariAlireza Zolanvari

35716

share|improve this question

edited Mar 20 at 11:07

Alireza Zolanvari

asked Mar 20 at 9:52

Alireza ZolanvariAlireza Zolanvari

35716

asked Mar 20 at 9:52

Alireza ZolanvariAlireza Zolanvari

35716

asked Mar 20 at 9:52

Alireza ZolanvariAlireza Zolanvari

35716

1 Answer
1

active

oldest

votes

1

$begingroup$

For unlabeled data-sets, unsupervised anomaly detectors can be compared either subjectively or objectively.

1. 
   

   Subjective comparison: based on our domain-knowledge and by using some visualizations and statistics, we can compare two detectors and select the one that outputs better anomalies subjectively.

   
   
   
   1. Here is a well-cited survey on unsupervised anomaly detectors that compares the algorithms on labeled data-sets (with known, domain-specific outliers) using AUC, and concludes that local detectors (such as LOF,
      COF, INFLO and LoOP) are not good candidates for global anomaly detection:
      2016 A Comparative Evaluation of Unsupervised Anomaly Detection Algorithms for Multivariate Data
      
   
   



2. Objective comparison: possible in theory, impossible in practice.

Requirements for objective comparison:

1. Anomaly definition: $x$ is an anomaly if $P(x)< t$ for some threshold $t$,




2. Anomaly detector requirement: $D$ is an anomaly detector if for every detected $x$, $P(x)< t$,




3. Comparing anomalies: $x_1$ is more anomalous than $x_2$ if $P(x_1)<P(x_2)$ or equivalently $r(x_1, x_2) = P(x_1) / P(x_2) < 1$,




4. Comparing anomaly detectors: proposal $x_1$ from detector $D_1$ is better than $x_2$ from $D_2$ if $r(x_1, x_2) < 1$,

As you can see, for qualification and comparison of two detectors we need to know $P(x)$ or at least $r(x_1, x_2)$. But if we know these quantities (which act as a judge $J$) or at least a close enough estimation of them, we have a better anomaly detector $J$ and can throw $D_1$ and $D_2$ away! We plug any observation $x$ or pair of observations $x_1$ and $x_2$ into $J$ and check which one is an anomaly or which one is more anomalous, done! So it is impossible to compare two anomaly detectors objectively unless we have a better anomaly detector (judge). So we should use a subjective comparison.

share|improve this answer

edited Mar 20 at 14:21

answered Mar 20 at 10:57

EsmailianEsmailian

1,696115

$endgroup$

• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  Please check the question update. Each sample has about 40 features and subjective comparison is not very practical.
  $endgroup$
  – Alireza Zolanvari
  Mar 20 at 11:04
  

  
  
  
  

add a comment | 

Your Answer

StackExchange.ifUsing("editor", function ()
return StackExchange.using("mathjaxEditing", function ()
StackExchange.MarkdownEditor.creationCallbacks.add(function (editor, postfix)
StackExchange.mathjaxEditing.prepareWmdForMathJax(editor, postfix, [["$", "$"], ["\\(","\\)"]]);
);
);
, "mathjax-editing");

StackExchange.ready(function()
var channelOptions =
tags: "".split(" "),
id: "557"
;
initTagRenderer("".split(" "), "".split(" "), channelOptions);

StackExchange.using("externalEditor", function()
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled)
StackExchange.using("snippets", function()
createEditor();
);

else
createEditor();

);

function createEditor()
StackExchange.prepareEditor(
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: false,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: null,
bindNavPrevention: true,
postfix: "",
imageUploader:
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
,
onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
);



);

draft saved

draft discarded

Sign up or log in

StackExchange.ready(function ()
StackExchange.helpers.onClickDraftSave('#login-link');
);

Sign up using Google

Sign up using Facebook

Sign up using Email and Password

Post as a guest

Name

Email

Required, but never shown

StackExchange.ready(
function ()
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fdatascience.stackexchange.com%2fquestions%2f47658%2fhow-to-compare-the-performance-of-two-unsupervised-algorithms-on-same-data-set%23new-answer', 'question_page');

);

Post as a guest

Name

Email

Required, but never shown

1

$begingroup$

For unlabeled data-sets, unsupervised anomaly detectors can be compared either subjectively or objectively.

1. 
   

   Subjective comparison: based on our domain-knowledge and by using some visualizations and statistics, we can compare two detectors and select the one that outputs better anomalies subjectively.

   
   
   
   1. Here is a well-cited survey on unsupervised anomaly detectors that compares the algorithms on labeled data-sets (with known, domain-specific outliers) using AUC, and concludes that local detectors (such as LOF,
      COF, INFLO and LoOP) are not good candidates for global anomaly detection:
      2016 A Comparative Evaluation of Unsupervised Anomaly Detection Algorithms for Multivariate Data
      
   
   



2. Objective comparison: possible in theory, impossible in practice.

Requirements for objective comparison:

1. Anomaly definition: $x$ is an anomaly if $P(x)< t$ for some threshold $t$,




2. Anomaly detector requirement: $D$ is an anomaly detector if for every detected $x$, $P(x)< t$,




3. Comparing anomalies: $x_1$ is more anomalous than $x_2$ if $P(x_1)<P(x_2)$ or equivalently $r(x_1, x_2) = P(x_1) / P(x_2) < 1$,




4. Comparing anomaly detectors: proposal $x_1$ from detector $D_1$ is better than $x_2$ from $D_2$ if $r(x_1, x_2) < 1$,

As you can see, for qualification and comparison of two detectors we need to know $P(x)$ or at least $r(x_1, x_2)$. But if we know these quantities (which act as a judge $J$) or at least a close enough estimation of them, we have a better anomaly detector $J$ and can throw $D_1$ and $D_2$ away! We plug any observation $x$ or pair of observations $x_1$ and $x_2$ into $J$ and check which one is an anomaly or which one is more anomalous, done! So it is impossible to compare two anomaly detectors objectively unless we have a better anomaly detector (judge). So we should use a subjective comparison.

share|improve this answer

edited Mar 20 at 14:21

answered Mar 20 at 10:57

EsmailianEsmailian

1,696115

$endgroup$

• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  Please check the question update. Each sample has about 40 features and subjective comparison is not very practical.
  $endgroup$
  – Alireza Zolanvari
  Mar 20 at 11:04
  

  
  
  
  

add a comment | 

1

$begingroup$

For unlabeled data-sets, unsupervised anomaly detectors can be compared either subjectively or objectively.

1. 
   

   Subjective comparison: based on our domain-knowledge and by using some visualizations and statistics, we can compare two detectors and select the one that outputs better anomalies subjectively.

   
   
   
   1. Here is a well-cited survey on unsupervised anomaly detectors that compares the algorithms on labeled data-sets (with known, domain-specific outliers) using AUC, and concludes that local detectors (such as LOF,
      COF, INFLO and LoOP) are not good candidates for global anomaly detection:
      2016 A Comparative Evaluation of Unsupervised Anomaly Detection Algorithms for Multivariate Data
      
   
   



2. Objective comparison: possible in theory, impossible in practice.

Requirements for objective comparison:

1. Anomaly definition: $x$ is an anomaly if $P(x)< t$ for some threshold $t$,




2. Anomaly detector requirement: $D$ is an anomaly detector if for every detected $x$, $P(x)< t$,




3. Comparing anomalies: $x_1$ is more anomalous than $x_2$ if $P(x_1)<P(x_2)$ or equivalently $r(x_1, x_2) = P(x_1) / P(x_2) < 1$,




4. Comparing anomaly detectors: proposal $x_1$ from detector $D_1$ is better than $x_2$ from $D_2$ if $r(x_1, x_2) < 1$,

As you can see, for qualification and comparison of two detectors we need to know $P(x)$ or at least $r(x_1, x_2)$. But if we know these quantities (which act as a judge $J$) or at least a close enough estimation of them, we have a better anomaly detector $J$ and can throw $D_1$ and $D_2$ away! We plug any observation $x$ or pair of observations $x_1$ and $x_2$ into $J$ and check which one is an anomaly or which one is more anomalous, done! So it is impossible to compare two anomaly detectors objectively unless we have a better anomaly detector (judge). So we should use a subjective comparison.

share|improve this answer

edited Mar 20 at 14:21

answered Mar 20 at 10:57

EsmailianEsmailian

1,696115

$endgroup$

• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  Please check the question update. Each sample has about 40 features and subjective comparison is not very practical.
  $endgroup$
  – Alireza Zolanvari
  Mar 20 at 11:04
  

  
  
  
  

add a comment | 

$begingroup$

For unlabeled data-sets, unsupervised anomaly detectors can be compared either subjectively or objectively.

1. 
   

   Subjective comparison: based on our domain-knowledge and by using some visualizations and statistics, we can compare two detectors and select the one that outputs better anomalies subjectively.

   
   
   
   1. Here is a well-cited survey on unsupervised anomaly detectors that compares the algorithms on labeled data-sets (with known, domain-specific outliers) using AUC, and concludes that local detectors (such as LOF,
      COF, INFLO and LoOP) are not good candidates for global anomaly detection:
      2016 A Comparative Evaluation of Unsupervised Anomaly Detection Algorithms for Multivariate Data
      
   
   



2. Objective comparison: possible in theory, impossible in practice.

Requirements for objective comparison:

1. Anomaly definition: $x$ is an anomaly if $P(x)< t$ for some threshold $t$,




2. Anomaly detector requirement: $D$ is an anomaly detector if for every detected $x$, $P(x)< t$,




3. Comparing anomalies: $x_1$ is more anomalous than $x_2$ if $P(x_1)<P(x_2)$ or equivalently $r(x_1, x_2) = P(x_1) / P(x_2) < 1$,




4. Comparing anomaly detectors: proposal $x_1$ from detector $D_1$ is better than $x_2$ from $D_2$ if $r(x_1, x_2) < 1$,

As you can see, for qualification and comparison of two detectors we need to know $P(x)$ or at least $r(x_1, x_2)$. But if we know these quantities (which act as a judge $J$) or at least a close enough estimation of them, we have a better anomaly detector $J$ and can throw $D_1$ and $D_2$ away! We plug any observation $x$ or pair of observations $x_1$ and $x_2$ into $J$ and check which one is an anomaly or which one is more anomalous, done! So it is impossible to compare two anomaly detectors objectively unless we have a better anomaly detector (judge). So we should use a subjective comparison.

share|improve this answer

edited Mar 20 at 14:21

answered Mar 20 at 10:57

EsmailianEsmailian

1,696115

$endgroup$

share|improve this answer

edited Mar 20 at 14:21

answered Mar 20 at 10:57

EsmailianEsmailian

1,696115

answered Mar 20 at 10:57

EsmailianEsmailian

1,696115

answered Mar 20 at 10:57

EsmailianEsmailian

1,696115