How to gather data of same class, select N rows, calculate mean in CSV file

1 次查看（过去 30 天）
Kong
Kong 2020-3-17
编辑： Hiro Yoshino 2020-3-17
Hello.
I have a data (90 x 2857), column 2857 is a label(class)
I want to gather data of the same class, select N rows, calculate mean.
I made the below code, but I want to reduce the code using for loop
Thank you.
clear all
close all
% read csv file
final = csvread('outfile.csv');
% split data and label
predictors = final(:, 1:end-1);
labels = final(:, end);
predictors = normalize(predictors, 2);
predictors_train = predictors(1:80, :); % rows 1 to 80
predictors_test = predictors(81:90, :); % rows 81 to 90
labels_train = labels(1:80, :); % rows 1 to 80
labels_test = labels(81:90, :); % rows 81 to 90
% gather data of same lable
wave2_support= [predictors_train(1,:); predictors_train(11,:); predictors_train(21,:);
predictors_train(31,:); predictors_train(41,:); predictors_train(51,:);
predictors_train(61,:); predictors_train(71,:)];
wave1_support= [predictors_train(2,:); predictors_train(12,:); predictors_train(22,:);
predictors_train(32,:); predictors_train(42,:); predictors_train(52,:);
predictors_train(62,:); predictors_train(72,:)];
walk_support= [predictors_train(3,:); predictors_train(13,:); predictors_train(23,:);
predictors_train(33,:); predictors_train(43,:); predictors_train(53,:);
predictors_train(63,:); predictors_train(73,:)];
skip_support= [predictors_train(4,:); predictors_train(14,:); predictors_train(24,:);
predictors_train(34,:); predictors_train(44,:); predictors_train(54,:);
predictors_train(64,:); predictors_train(74,:)];
side_support= [predictors_train(5,:); predictors_train(15,:); predictors_train(25,:);
predictors_train(35,:); predictors_train(45,:); predictors_train(55,:);
predictors_train(65,:); predictors_train(75,:)];
run_support= [predictors_train(6,:); predictors_train(16,:); predictors_train(26,:);
predictors_train(36,:); predictors_train(46,:); predictors_train(56,:);
predictors_train(66,:); predictors_train(76,:)];
pjump_support= [predictors_train(7,:); predictors_train(17,:); predictors_train(27,:);
predictors_train(37,:); predictors_train(47,:); predictors_train(57,:);
predictors_train(67,:); predictors_train(77,:)];
jump_support= [predictors_train(8,:); predictors_train(18,:); predictors_train(28,:);
predictors_train(38,:); predictors_train(48,:); predictors_train(58,:);
predictors_train(68,:); predictors_train(78,:)];
jack_support= [predictors_train(9,:); predictors_train(19,:); predictors_train(29,:);
predictors_train(39,:); predictors_train(49,:); predictors_train(59,:);
predictors_train(69,:); predictors_train(79,:)];
bend_support= [predictors_train(10,:); predictors_train(20,:); predictors_train(30,:);
predictors_train(40,:); predictors_train(50,:); predictors_train(60,:);
predictors_train(70,:); predictors_train(80,:)];
% randomly select 5 rows of 8 data in same label
N=5;
c = randperm(size(wave2_support,1)); c = c(1:N);
wave2=wave2_support(c,:); % wave2 matrix
c = randperm(size(wave1_support,1)); c = c(1:N);
wave1=wave1_support(c,:); % wave1 matrix
c = randperm(size(walk_support,1)); c = c(1:N);
walk=walk_support(c,:); % walk matrix
c = randperm(size(skip_support,1)); c = c(1:N);
skip=skip_support(c,:); % skip matrix
c = randperm(size(side_support,1)); c = c(1:N);
side=side_support(c,:); % side matrix
c = randperm(size(run_support,1)); c = c(1:N);
run=run_support(c,:); % run matrix
c = randperm(size(pjump_support,1)); c = c(1:N);
pjump=pjump_support(c,:); % pjump matrix
c = randperm(size(jump_support,1)); c = c(1:N);
jump=jump_support(c,:); % jump matrix
c = randperm(size(jack_support,1)); c = c(1:N);
jack=jack_support(c,:); % jack matrix
c = randperm(size(bend_support,1)); c = c(1:N);
bend=bend_support(c,:); % bend matrix
% calculate mean of each support data
wave2_mean = mean(wave2,1);
wave1_mean = mean(wave1,1);
walk_mean = mean(walk,1);
skip_mean = mean(skip,1);
side_mean = mean(side,1);
run_mean = mean(run,1);
pjump_mean = mean(pjump,1);
jump_mean = mean(jump,1);
jack_mean = mean(jack,1);
bend_mean = mean(bend,1);
% select query (1 row)
N=1; % 1 of row randomly selected
c = randperm(size(predictors_test,1)); c = c(1:N);
query=predictors_test(c,:); % output matrix
% calculate mean of each support
euclidean_bend = pdist2(bend_mean,query, 'euclidean');
euclidean_jack = pdist2(jack_mean,query, 'euclidean');
euclidean_jump = pdist2(jump_mean,query, 'euclidean');
euclidean_pjump = pdist2(pjump_mean,query, 'euclidean');
euclidean_run = pdist2(run_mean,query, 'euclidean');
euclidean_side = pdist2(side_mean,query, 'euclidean');
euclidean_skip = pdist2(skip_mean,query, 'euclidean');
euclidean_walk = pdist2(walk_mean,query, 'euclidean');
euclidean_wave1 = pdist2(wave1_mean,query, 'euclidean');
euclidean_wave2 = pdist2(wave2_mean,query, 'euclidean');
% softmax
n = [euclidean_bend; euclidean_jack; euclidean_jump; euclidean_pjump;
euclidean_run; euclidean_side; euclidean_skip;
euclidean_walk; euclidean_wave1; euclidean_wave2];

请先登录，再回答此问题。

采纳的回答

Hiro Yoshino
Hiro Yoshino 2020-3-17
编辑：Hiro Yoshino 2020-3-17
I would suggest you should use table for data analytics.
The following code may help you reduce the number of lines.
up to the end line, the data is sorted by the lable.
data = readtable("outfile.csv");
data.Properties.VariableNames{end} = 'label';
data{:,1:end-1} = normalize(data{:,1:end-1},1); % I belive this should be "1"
data = sortrows(data,'label'); % The data are sorted by the lables
wave2_support = data(:,1:9);
table:
https://jp.mathworks.com/help/matlab/tables.html?s_tid=CRUX_lftnav
if you want to split the data into TEST and TRAIN, then you should try cvpatition:
https://jp.mathworks.com/help/stats/cvpartition.html
This function can maintain the ratio of the labels as it is after spliting the data into two.

更多回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Dimensionality Reduction and Feature Extraction 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by