{ "id": 11221, "title": "Reading data in MATLAB", "url": "/competitions/avazu-ctr-prediction/discussion/11221", "author_name": "", "post_date": "2014-12-13T01:26:23.597Z", "votes": null, "comment_count": 4, "views": 1631, "content": "

How to read this data in MATLAB? csvread doesn't work!

", "messages": [ { "id": "60126", "postDate": "12/13/2014 01:26:23", "content": "

How to read this data in MATLAB? csvread doesn't work!

", "rawMarkdown": "", "votes": null }, { "id": "60140", "postDate": "12/13/2014 10:42:15", "content": "

just try these MATLAB API :

\n

fid = fopen('path/filename.csv') ;

\n

dcells = textscan(fid, 'your-format-string' , 'Delimiter', ',', 'Headerlines', 1  ) ;

\n

fclose(fid) ;

\n

Them ,all data have been load in dcells .

\n

good luck ~

", "rawMarkdown": "", "votes": null }, { "id": "60165", "postDate": "12/13/2014 23:09:30", "content": "

Thanks! 

", "rawMarkdown": "", "votes": null }, { "id": "60166", "postDate": "12/13/2014 23:27:26", "content": "

Or better yet, use the new datastore feature in R2014b. See the example code below. 

\n

Cheers!

\n\n

%% Avazu Example Datastore Code

\n

ftrain = 'train.csv';

\n

% Create datastore object to perform partial data reads

\n

ds = datastore(ftrain);

\n

% Define data formats for each variable

\n

% variables removed from new dataset: device_os, device_make, device_geo_country

\n

ds.SelectedFormats = {'%q'... % id

\n

'%d'... % click

\n

'%d'... % hour

\n

'%C'... % C1

\n

'%C'... % banner_pos

\n

'%C'... % site_id

\n

'%C'... % site_domain

\n

'%C'... % site_category

\n

'%C'... % app_id

\n

'%C'... % app_domain

\n

'%C'... % app_category

\n

'%q'... % device_id - ignore

\n

'%q'... % device_ip - ignore

\n

'%C'... % device_model

\n

'%C'... % device_type

\n

'%C'... % device_conn_type

\n

'%C'... % C14

\n

'%C'... % C15

\n

'%C'... % C16

\n

'%C'... % C17

\n

'%C'... % C18

\n

'%C'... % C19

\n

'%C'... % C20

\n

'%C'}; % C21

\n\n

%% Read in all of the click variable

\n

ds.SelectedVariableNames = 'click';

\n

click = readall(ds);

\n

click = logical(click.click);

\n

%% Partial reads to sample

\n

% Read in first row to initialize table variable

\n

ds.RowsPerRead = 1;

\n

data = read(ds);

\n

% Reset number of rows per read

\n

ds.RowsPerRead = 20000;

\n

% Read in a block and process

\n

while hasdata(ds)

\n

tempdata = read(ds);

\n

% Pull random sample without replacement

\n

tempdata = datasample(tempdata,100,'Replace',false);

\n\n

data = [data; tempdata];

\n

end

", "rawMarkdown": "", "votes": null }, { "id": "60168", "postDate": "12/13/2014 23:47:26", "content": "

Textscan doesnt seem to work because data doesn't fit in memory. Also, datastore isn't present in R2014a, which is the version of MATLAB I am using.

\n

Any solution to this data problem?

", "rawMarkdown": "", "votes": null } ], "comments": [ { "id": 60140, "author_name": "lemoner", "author_url": "", "post_date": "12/13/2014 10:42:15", "content": "

just try these MATLAB API :

\n

fid = fopen('path/filename.csv') ;

\n

dcells = textscan(fid, 'your-format-string' , 'Delimiter', ',', 'Headerlines', 1  ) ;

\n

fclose(fid) ;

\n

Them ,all data have been load in dcells .

\n

good luck ~

", "votes": null, "replies": [] }, { "id": 60165, "author_name": "anujamkelkar", "author_url": "", "post_date": "12/13/2014 23:09:30", "content": "

Thanks! 

", "votes": null, "replies": [] }, { "id": 60166, "author_name": "toddschultz", "author_url": "", "post_date": "12/13/2014 23:27:26", "content": "

Or better yet, use the new datastore feature in R2014b. See the example code below. 

\n

Cheers!

\n\n

%% Avazu Example Datastore Code

\n

ftrain = 'train.csv';

\n

% Create datastore object to perform partial data reads

\n

ds = datastore(ftrain);

\n

% Define data formats for each variable

\n

% variables removed from new dataset: device_os, device_make, device_geo_country

\n

ds.SelectedFormats = {'%q'... % id

\n

'%d'... % click

\n

'%d'... % hour

\n

'%C'... % C1

\n

'%C'... % banner_pos

\n

'%C'... % site_id

\n

'%C'... % site_domain

\n

'%C'... % site_category

\n

'%C'... % app_id

\n

'%C'... % app_domain

\n

'%C'... % app_category

\n

'%q'... % device_id - ignore

\n

'%q'... % device_ip - ignore

\n

'%C'... % device_model

\n

'%C'... % device_type

\n

'%C'... % device_conn_type

\n

'%C'... % C14

\n

'%C'... % C15

\n

'%C'... % C16

\n

'%C'... % C17

\n

'%C'... % C18

\n

'%C'... % C19

\n

'%C'... % C20

\n

'%C'}; % C21

\n\n

%% Read in all of the click variable

\n

ds.SelectedVariableNames = 'click';

\n

click = readall(ds);

\n

click = logical(click.click);

\n

%% Partial reads to sample

\n

% Read in first row to initialize table variable

\n

ds.RowsPerRead = 1;

\n

data = read(ds);

\n

% Reset number of rows per read

\n

ds.RowsPerRead = 20000;

\n

% Read in a block and process

\n

while hasdata(ds)

\n

tempdata = read(ds);

\n

% Pull random sample without replacement

\n

tempdata = datasample(tempdata,100,'Replace',false);

\n\n

data = [data; tempdata];

\n

end

", "votes": null, "replies": [] }, { "id": 60168, "author_name": "anujamkelkar", "author_url": "", "post_date": "12/13/2014 23:47:26", "content": "

Textscan doesnt seem to work because data doesn't fit in memory. Also, datastore isn't present in R2014a, which is the version of MATLAB I am using.

\n

Any solution to this data problem?

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "60126": "", "60140": "", "60165": "", "60166": "", "60168": "" }, "source": "meta" }