Using barh to plot top 10 values in Feature Selection

Question

Nina Perf 2021-9-6

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1447889-using-barh-to-plot-top-10-values-in-feature-selection

编辑： Nina Perf 2021-9-6

采纳的回答： Star Strider

在 MATLAB Online 中打开

Hello,

I did Feature Selection of 700 features.

I want to bar plot only the top 5 (with highest predictor rank).

Can you please help?

Thank you in advance.

[idx,scores] = fscmrmr(t_FeaturesData(:,2:700),t.score);
barh(scores(idx))
xlabel('Predictor rank')
ylabel('Predictor importance score')

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

Star Strider 2021-9-6

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1447889-using-barh-to-plot-top-10-values-in-feature-selection#answer_781954

I have no idea what the data are, however the maxk function could do what you want.

.

2 个评论
显示无隐藏无

Nina Perf 2021-9-6

在 MATLAB Online 中打开

[idx,scores] = fscmrmr(t_FeaturesData(:,2:700),t.score);
barh(scores(idx))
xlabel('Predictor rank')
ylabel('Predictor importance score')

Thank you so much for the feedback.

Do you know how to use with this? I want to plot a similar barh.

Star Strider 2021-9-6

My pleasure!

I did not see that part of the post earlier.

Looking through the documentation for the fscmrmr funciton (that I have no experience with), it would seem that the 5 most important predictors are given by the first 5 elements of idx, as described in the example in the documentation section Rank Predictors by Importance . That would appear to do exactly what you describe. If it does not, use maxk on the relevant variable, that most likely being ‘score’.

.

请先登录，再进行评论。