I have a question, related to parallel work on python
How I can use Processers =1,2,3... on k nearest neighbor algorithm when K=1, 2, 3,.. to find the change in time spent, speedup, and efficiency.
What is the appropriate code for that?
$\begingroup$
$\endgroup$
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
AFAIK KNN take time only if size of the data is very huge. Meanwhile normal GridSearchCV itself works well with more K values, Refer this site - https://machinelearningknowledge.ai/knn-classifier-in-sklearn-using-gridsearchcv-with-example/.
Still you want to run in parallel please create a method in which you fit the KNN by accepting the K value as parameter in the method.
Create subprocess for each K value in the main method and call the KNN method by passing the K value.
Code will be more complex if you write like this since we need to write logic to compare the accuracy of each fit w.r.t cross validation etc.
