计算两对 X 和 y 之间的相似性的最佳做法是什么

Question

I have some values about one element. For example, element1: values1, values2 . For each element, I need to calculate the 'score' for a given number of features. Imagine that we have one feature that is represented as:

• An high score for the feature1 is given by an high score of value1 And a low score of value2.

So If I suppose that to an high score of value1 (1) And a low score of value2 (0) correspond an high score of 'feature1', what is the best practice to calculate the score of feature1 given as value1 And value2 two different scores? (For example value1=0.7, value=0.2). I use Python as programming language, And I prefer to use sklearn ad module but every solution that fits well is accepted.

Answer 1

1. First normalize your data. One type of normalization is to make your values1, values2 fit between the range [0,1].
2. Suppose the average 2-value characterization of the feature1 based on the normalized data is (.7, .2). For any new 2-values (x,y) compute the distance between (x,y) and (.7,.2)

When computing distance in machine learning, the sqrt component is usually not calculated.

dist^2 = (x-.7)^2 + (y-.2)^2

You might also be interested in calculating the error of a 2-value (x,y) wrt to (.7,.2) and can look into categorical cross entropy.