Skip to main content
Cross Validated

Return to Question

Notice removed Draw attention by Tomek Tarczynski
Bounty Ended with Zhubarb's answer chosen by Tomek Tarczynski
deleted 4 characters in body; edited title
Source Link
amoeba
amoeba
  • 107.3k
  • 36
  • 321
  • 346

Practical knowledge from What classification algorithm should one use after seeing that t-SNE separates classes well?

LetsLet's assume we have a classification problem and at first we want to get some insight from the data and we do t-SNE. The result of t-SNE separates classes very well. This implies that it is possible to build classification model that will also separatesseparate classes very well (if t-SNE doesn't separate well then it doesn't imply much).

Knowing that t-SNE focuses on local structure and that it can separate classes well: What are the classification algorithms that should work well on thatthis problem? Scikit suggests SVM with a Gaussian RBF kernel, but what are the others?

Practical knowledge from t-SNE

Lets assume we have a classification problem and at first we want to get some insight from the data and we do t-SNE. The result of t-SNE separates classes very well. This implies that it is possible to build classification model that will also separates classes very well (if t-SNE doesn't separate well then it doesn't imply much).

Knowing that t-SNE focuses on local structure and that it can separate classes well: What are the classification algorithms that should work well on that problem? Scikit suggests SVM with a Gaussian RBF kernel, but what are the others?

What classification algorithm should one use after seeing that t-SNE separates classes well?

Let's assume we have a classification problem and at first we want to get some insight from the data and we do t-SNE. The result of t-SNE separates classes very well. This implies that it is possible to build classification model that will also separate classes very well (if t-SNE doesn't separate well then it doesn't imply much).

Knowing that t-SNE focuses on local structure and that it can separate classes well: What are classification algorithms that should work well on this problem? Scikit suggests SVM with a Gaussian RBF kernel, but what are the others?

Notice added Draw attention by Tomek Tarczynski
Bounty Started worth 50 reputation by Tomek Tarczynski
Tweeted twitter.com/StackStats/status/850426982101897217
Source Link
Tomek Tarczynski
Tomek Tarczynski
  • 4k
  • 7
  • 35
  • 40

Practical knowledge from t-SNE

Lets assume we have a classification problem and at first we want to get some insight from the data and we do t-SNE. The result of t-SNE separates classes very well. This implies that it is possible to build classification model that will also separates classes very well (if t-SNE doesn't separate well then it doesn't imply much).

Knowing that t-SNE focuses on local structure and that it can separate classes well: What are the classification algorithms that should work well on that problem? Scikit suggests SVM with a Gaussian RBF kernel, but what are the others?