[PYTHON] Can you challenge Kaggle with just your iPad?

Personally, I often do something on the iPad, and I've been doing a lot of trial and error to see if I can code it. I also created an environment for posting Kaggle in the flow of forcibly building a coding environment with theia.

"Let's make a survival prediction model for Kaggle Titanic passengers without using Python" Since I was able to participate in Kaggle with GUI, can I use iPad? I thought and tried it. By the way, this article is also written from the iPad.

Participate in competitions and download files

I will participate in the Titanic competition from my iPad. Log in to Kaggle from your iPad to download the dataset.

• By the way, I have omitted participation in the competition, but if you are interested, please refer to the article above.

When you open the "File" app on the iPad, the downloaded file is in the download folder, so unzip it.

Make a prediction model with VARISTA

Go back to Safari and open VARISTA. VARISTA is here.

After opening VARISTA, create a new project.

For the file, select "train.csv" from the folder you downloaded earlier.

As with the previous article, the learning settings are level 3, the ratio of validation data is 5%, the number of cross-validation divisions is 10, and the random seed is 123, and learning is started.

When you're done training, select Predict with this model from the model and select the columns you want to delete.

The items required for Submission are PassengerId and Predicted, so delete the others.

After that, change the format of the column to be predicted to a flag.

Then upload "test.csv" as a prediction file.

Once the prediction is complete, download the file and finally submit it to Kaggle.

Click Kaggle's Submi Prediction and select the file you downloaded from Browse.

All you have to do is Make Submission and you're done.

Since it is not possible to concatenate columns, it is necessary to use numbers, but since you can see a simple breakdown of the data, it seems that you can use it easily for the time being. It may be good to see the contents of the data quickly.

VARISTA is nice because it looks cool.

reference

VARISTA: https://www.varista.ai Kaggle: https://www.kaggle.com