Task 2 -- Mandatory Tasks: Difference between revisions

From Master of Neuroscience Wiki
Created page with "Errors are an option… === 2 === Basic:<div class="figure"> File:31 0.png </div>Scaled by divison max() for every individual frequency band: File:31 1.png<div class="figure"> </div> == 3 == without preparing the data via /= std:<div class="figure"> File:31 2.png </div>with equalizing the power via /= std (obviously not the best idea in this case):<div class="figure"> File:31 3.png </div> == 4 == Phase Coherence<div class="figure"> File:31 4.png <..."
 
 
(One intermediate revision by the same user not shown)
Line 4: Line 4:
Basic:<div class="figure">
Basic:<div class="figure">
[[File:31 0.png]]
[[File:31 0.png]]
</div>Scaled by divison max() for every individual frequency band:
</div>Scaled by divison max() for every individual frequency band:[[File:31 1.png]]<div class="figure">
 
[[File:31 1.png]]<div class="figure">
</div>
</div>


Line 33: Line 31:
[[File:31 11.png]]
[[File:31 11.png]]
</div>Spectral Coherence<div class="figure">
</div>Spectral Coherence<div class="figure">
[[File:Image8.png|image8.png]]
[[File:31 8.png]]
</div><div class="figure">
</div><div class="figure">
[[File:Image9.png|image9.png]]
[[File:31 9.png]]
</div>
</div>


== 6 ==
== 6 ==
{: .topic-optional} Don’t normalize the time series!
'''Don’t normalize the time series!'''
 
 


i.e. don’t do something like this:<syntaxhighlight lang="python">data -= data.mean(axis=1, keepdims=True)
i.e. don’t do something like this:<syntaxhighlight lang="python">data -= data.mean(axis=1, keepdims=True)
data /= data.std(axis=1, keepdims=True)</syntaxhighlight>Otherwise you will not classify anything.<div class="figure">
data /= data.std(axis=1, keepdims=True)</syntaxhighlight>Otherwise you will not classify anything.<div class="figure">
[[File:Image12.png|image12.png]]
[[File:31 12.png]]
</div>Scaled by divison max() for every individual frequency band (Bad times happen):<div class="figure">
</div>Scaled by divison max() for every individual frequency band (Bad times happen):<div class="figure">
[[File:Image13.png|image13.png]]
[[File:31 13.png]]
</div>
</div>


== 7 ==
== 7 ==
{: .topic-optional} Don’t normalize the time series!
'''Don’t normalize the time series!'''
 
 


i.e. don’t do something like this:<syntaxhighlight lang="python">data -= data.mean(axis=1, keepdims=True)
i.e. don’t do something like this:<syntaxhighlight lang="python">data -= data.mean(axis=1, keepdims=True)
data /= data.std(axis=1, keepdims=True)</syntaxhighlight>Otherwise you will not classify anything.<div class="figure">
data /= data.std(axis=1, keepdims=True)</syntaxhighlight>Otherwise you will not classify anything.<div class="figure">
[[File:Image14.png|image14.png]]
[[File:31 14.png]]
</div>
</div>

Latest revision as of 13:20, 23 October 2025

Errors are an option…

2

Basic:

Scaled by divison max() for every individual frequency band:

3

without preparing the data via /= std:

with equalizing the power via /= std (obviously not the best idea in this case):

4

Phase Coherence

Spectral Coherence

5

Phase Coherence

Spectral Coherence

6

Don’t normalize the time series!


i.e. don’t do something like this:

data -= data.mean(axis=1, keepdims=True)
data /= data.std(axis=1, keepdims=True)

Otherwise you will not classify anything.

Scaled by divison max() for every individual frequency band (Bad times happen):

7

Don’t normalize the time series!


i.e. don’t do something like this:

data -= data.mean(axis=1, keepdims=True)
data /= data.std(axis=1, keepdims=True)

Otherwise you will not classify anything.