Top ranking features using fscmrmr() have 0 for scores

Question

Tyler Hoang 2022-3-4

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1663319-top-ranking-features-using-fscmrmr-have-0-for-scores

回答： Prince Kumar 2022-3-31

Hi,

I used fscmrmr() on a dataset of 100 observations by 1000 features. The top 100 features have scores 0 where as the remaining features have some scores > 0. Is this normal?

Assuming the idx and scores below were already ranked.

[idx,scores] = fscmrmr(X,Y);

Many thanks

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Prince Kumar 2022-3-31

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1663319-top-ranking-features-using-fscmrmr-have-0-for-scores#answer_931259

在 MATLAB Online 中打开

Hi,

Please have a look at the following example from fscmrmr for better understanding.

load ionosphere

[idx,scores] = fscmrmr(X,Y);

bar(scores(idx))

xlabel('Predictor rank')

ylabel('Predictor importance score')

One important thing to note is that idx and scores are in sorted order.

scores
scores = 1×34
    0.0662         0    0.0214    0.0756    0.2905    0.0068    0.0345    0.0085    0.0128    0.0063    0.0023    0.0274    0.0079    0.0007    0.0145    0.0014    0.0018         0    0.0073         0    0.0141         0    0.0066    0.0310    0.0086         0    0.0098         0    0.0280    0.0072
idx
idx = 1×34
     5     4     1     7    24    29    12     3    34    33    15    21     9    32    31    27    25     8    13    19    30     6    23    10    11    17    16    14     2    18