How do I download .csv dataset directly from Kaggle with Matlab code.

Hello, I found a dataset on Kaggle which I want to use for my Machine learning project. However, I'm required to download the dataset from the code itself. I tried this:

url1 = 'https://www.kaggle.com/competitions/mrs-spring-2023-battery-prediction-challenge/data?select=example_submission.csv';
url2 = 'https://www.kaggle.com/competitions/mrs-spring-2023-battery-prediction-challenge/data?select=test_data.csv';
url3 = 'https://www.kaggle.com/competitions/mrs-spring-2023-battery-prediction-challenge/data?select=training_data.csv';

example = websave('example_submission.csv',url1);
test = websave('test_data.csv',url2);
training = websave('training_data.csv',url3);

data_example = readtable("example_submission.csv");
data_test = readtable("test_data.csv");
data_train = readtable("training_data.csv");

I was able to download them, however instead of downloading the .csv files, it is downloading an HTML page that contains the Kaggle website.

[Star Strider]

You probably have to download those directly from the Kaggle site after signing in. It is apparently not possible to download them directly from that site through MATLAB.

That problem aside, the code to get them using websave should be:

example = websave('example_submission',url1);
test = websave('test_data',url2);
training = websave('training_data',url3);

data_example = readtable(example);
data_test = readtable(test);
data_train = readtable(training);

That actually works (I tested it) however it returns the HTML code rather than the file contents in the table arrays.

.

.

[Emil Buratovic]

Thank you for your effort.

Are there any other sites with similar datasets which I could use to download directly through code?

[Star Strider]

My pleasure!

Not that I am aware of. You probably need to search for the sort of information you want, since I am not certain what that is.

If you have access to a university library, search using those computers because the data could be in journals that have paywalls, and the university could have access to them. Once the data are downloaded, you can send the files to your home computer via e-mail. (That is the approach I have used in the past. The university library computers I am familiar with prohibit using SD cards and such because of the malware danger.)

Answer 1

[Harsh]

Hi there,

It seems you're looking to download datasets from Kaggle using MATLAB. Currently, this is not feasible using only MATLAB for now. However, if Python is available, you can follow the workaround below to accomplish this by leveraging Kaggle's API commands, which are conveniently available on each dataset's page.

To start, you'll need to install the Kaggle package via pip:

pip install kaggle

Once installed, you can utilize MATLAB's system commands to execute these API calls and save the datasets as .csv files. Here's a snippet of the script I've written for this purpose:

% Download the zip file from kaggle 
callSystemCmd('kaggle datasets download -d spypsc07/amazon-products') 

% Unzip the zip file 
callSystemCmd('tar -xf amazon-products.zip') 

Here, in the above code I have created a function “callSystemCmd”

%% Here’s the code for it  
function callSystemCmd(cmd) 
    % Execute the command 
    [status, cmdout] = system(cmd); 
    % Check if the command was successful 
    if status == 0 
        disp('Command executed successfully:'); 
        disp(cmdout); 
    else 
        disp('Command execution failed.'); 
    end 
end 

After running the script above, you should see the dataset files neatly stored in your directory.

I hope this helps, thanks!

[Emil Buratovic]

Thank you, I have tried that route myself but the task forbid me from using more than 1 program language.

I found completely new data set elsewhere and completed my task, however, your suggestion works,