[PYTHON] [See in the photo] How a kaggle beginner can rank up from "novice" to "Contributor" in 10 minutes.

Introduction

Nikkei xTECH Business AI Advent Calendar 2019 I will be in charge of the 4th day of the road to the AI dojo "Kaggle".

kaggle's voice depends on status

Users are ranked in kaggle. It's called Progression. It's like a progress level or a title.

Depending on this rank, the degree of attention to the posted script and the degree of influence of the remark will differ.

At the time of registration, the title ** novice ** will be given, and if the prescribed conditions are met, it will become ** Contributor **. If you want to reach a higher rank from there, you have to make a statement to collect good evaluations from others, win medals, and so on.

kaggle is a competitive competition, and if you just want to collect medals, you should not submit your own answer until the last minute, read the trial results of others, and post it after improving the accuracy. That alone will not make kaggle prosper. kaggle is also a place for technical exchange. Contribute to kaggle's prosperity by posting your own answers to prosperity and commentary on the answers. By posting useful information, you can collect good evaluations and raise your rank.

If you register first, it's boring to be a beginner ** novice **, so let's aim for ** Contributor **. I have summarized how to become a ** Contributor ** in an easy-to-understand manner.

Register immediately

kaggle Official

There is a blue registration button in the upper right. I'm comfortable, so I'm logged in with a google account.

Check Progression rank

To check the rank, hover your mouse over ... on the top tab and click Progression.

As you scroll down, you'll see the achievement conditions for each rank. At first, I think Novice has a check mark. Novice's achievement condition is registration, so it is checked at the time of registration.

Let's take a look at the next higher rank Consumer we are aiming for this time.

Check items to achieve

・ Update bio ・ Add a place to live ・ Add job title ・ Add organization ・ Perform SMS authentication ・ Run the script ・ Submit the result to one competition ・ Add one comment ・ Give one like rating

It has become. Let's do it while checking one by one.

update bio

What is bio in the first place? An abbreviation for biography, it is a place to write a biography, but some people write about themselves and hobbies.

At the time of initial registration, I think that the image of Mr. Tori is displayed in the part where the login button on the upper right was. Place your mouse over Mr. Tori and click "my profile" to move to your profile screen.

You can see bio under the medal. Click edit on the right to add a self-introduction and you're good to go. I think Progression is checked.

Added address / title / organization

There is a blue edit profile button in the upper right of the figure where bio was added earlier. Press here

You will be given a place to enter your information. Job title in occupation Organization belonging to organization Address to city Let's enter.

Now there are three more checks.

Also, let's do SMS authentication. Scroll down a little and you will find the SMS verification link. If it has not been authenticated, it will be Not verified.

Let's actually submit it to the competition

・ Run the script ・ Submit the result ・ Add a comment

We will achieve this. I think this is the most wally thing. Does it look like it's annoying? There is this article for those who feel.

First, there is a search mark to the left of the competition on the top tab.

Enter titanic here to search. You will find a competition called "titanic: Machine Learning from Disaster" with a trophy mark at the top. The "<>" mark below it is a script written by someone.

Let's choose the titanic competition.

There is a join competition with the blue button, so join.

Read and agree to the terms.

join has changed to a button called submit. You can now write and submit a script using the data. Let's press the button.

You will see a script (notebook) written by someone else. Let's create a new script from the new notebook button.

Select the language in which you want to analyze. Select notebook or script as the submission format. If you want to add a description or execute it partially, notebook Choose your favorite language. This time, let's use R notebook instead of python.

As a demo code, explanations such as loading a library and loading a file are written.

For the time being, delete everything here and type in as follows. Execution of the cell of notebook is shift + enter like jupyter.

The working directory for this notebook is in kaggle. The files are in "..input", although there are some differences depending on the competition. When I checked the file in the input, there was a folder called "titanic". If you look further inside titanic, you will find three files.

You can predict the test using the model trained on the train.

The test is missing a column for the expected variable "whether it survived". Let's create a simple model to predict this.

Submit the prediction result with the write.csv function. There are column names that must be entered in the forecast data. This time, the passenger ID is applicable. Also, write.csv will write the line name, which will cause an error. Let's turn it off. When you have finished making it, press the commit button on the upper right to submit it.

If there are no particular errors, it means that the script has been registered. Press the over view button and go to submit.

If you scroll down your script that you can see in orver view, you can see the csv for submission that you made in the output files part. I think you can confirm.

It is finally submitted.

When it reaches complete, it is complete. You can also check the score.

In addition, let's comment the explanation etc. on your script.

This is checked.

Use the last one like

For example, pressing discussion on the top tab discusses scripts and techniques that others have written. Let's take a look inside and like what was interesting.

The number 111 is attached to the upper right of the figure. If you think it's good, press the △ button to support it.

Congratulations

This is the rank up. The rank goes up because you have experienced how to use kaggle. It's a testament to understanding the process of not only registering and reading people's scripts, but also participating in competitions and submitting results yourself.

Thank you for your hard work. Now, it starts from here. Let's enjoy kaggle.