一生懸命英語で記事を書いたんですが、Qiitaさんが投稿させてくれないので途中で放置します。
ここは僕の下手くそな英語で書かれてます。
ただの駄文なので気にしないでください。
結論を書くと、僕の能力不足と知識不足とデータ不足です。
英語は小学生レベルなので察してください。
If you do not know AKB48 and AKB48 election, this page is for you.
This page is written in my terrible English even living in USA.
Introduction
I like Japanese Idol. Especially, AKB48 groups are the best. My favorite groups are STU48 and NGT48.
I want to predict next AKB's election (11th) using disclosed SHOWROOM data using machine learning( kind of AI ) technique.
This page is my record of process making the prediction model. The accuracy of prediction will be appard next year(2019)!
I am not a data scientist and quiet beginner of Machine Learning and Programing.
What is SHOWROOM?
I can not find any description page about SHOWROOM in English, but, simply, SHOWROOM is one of the online streaming service like Twitch and Twicas. Almost all member of AKB48 group members have a account for streaming and many members are streaming every day at their home.
Why SHOWROOM?
I like SHOWROOM. You can check the SHOWROOM from all over the world. This service is not restricted in Japan. Some recorded movie is on Youtube illegally.
It could be easy to find out you that there are many datas to evaluate how the idol's famous. For example, Follower number of Twitter and Instagram, How often showed in TV, and so on.
1.Twitter and Instagram
Follower number of Twitter is very good data for prediction and "some" group member has twitter and instagram accounts. Unfortunately, STU48, the sister group of AKB, members do not have twitter and instagram individual account. So, I can not predict any STU48 member if I use twitter or Instagram.
2.TV
This is very difficult to get data. I am not professional TV analyst.