1
$\begingroup$

I have a data set that looks like the following:

 Time        V1     V2    V3     ...    V40
13:00        0.44   0     0.33          0.55
13:01        0.55   0     0.34          0.52
13:02        0.58   1     0.20          0.58
.
.
.
15:01        0.57   0     0.24          0.70

Where V2 is the binary equivalent of on/off switches. Currently, I am still pre-processing my data and normalized the data-set from (0,1) using sklearn.preprocessing. I am wondering if applying dimensionality reduction/PCA to my dataset will affect the outcome of my model and whether if it is advisable to use it to process my data.

$\endgroup$

2 Answers 2

1
$\begingroup$

I am wondering if applying dimensionality reduction/PCA to my dataset will affect the outcome of my model and whether if it is advisable to use it to process my data.

In short - you don't know. You need to perform experiments, to check, if reducing dimensionality helps your models to perform better. There's little things you can state a priori. Generally, 40 features isn't that much, to observe dimensionality curse. One you obviously need to do, is to check your features correlation and check if any of the feature damages your output.

$\endgroup$
0
$\begingroup$

If you have multi dimensional data, it's very hard to visualize. PCA helps us in reducing the dimension of the data set by keeping maximum co-variance in top K Features. So PCA will help us in presenting the data in most of the cases.

So if you perform PCA, you will definitely lose some data and creating a model with less data, will definitely affect the model outcome.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.