I think the topic is very interesting- but there are some concerns about how well your data fits your needs to answer your research question. It seems that data has only the top 1000 influencers. Those are already well established. In order to find out how to get successful, you need observations on new influencers and the established ones. Your models would be valid for established channels, which could be already interesting- but you need to distinguish your analysis from the already existing kaggle notebooks which can get challenging.
I think the topic is very interesting- but there are some concerns about how well your data fits your needs to answer your research question. It seems that data has only the top 1000 influencers. Those are already well established. In order to find out how to get successful, you need observations on new influencers and the established ones. Your models would be valid for established channels, which could be already interesting- but you need to distinguish your analysis from the already existing kaggle notebooks which can get challenging.