00:01
Okay, so here we are asked what are the steps to performing k and n in our studio.
00:07
So k nearest neighbors is going to be a simple algorithm that stores all the available cases and then classifies the new cases by a majority of its k neighbors.
00:20
So this algorithm is going to segregate the unlabeled data points into well -defined groups.
00:25
So the requirements for k and n are going to be first.
00:28
Generally, k gets decided on the square root of the number of data points.
00:34
But a large k value has benefits, which is going to include reducing the variance due to the noisy data.
00:42
And the side effect here is being developed a bias to which the learner tends to ignore these smaller patterns, which may have useful insights.
00:53
And then two, data normalization.
00:55
So it is the transform of all the feature...