What I wanted to do

For a group of users who are moving comments to a specific channel, acquire which channel each user has registered. Is it possible to estimate the customer base by doing so? Assumption. If there are multiple users who have subscribed to the same channel, it is possible that the customer base is those who are interested in the topics of that channel. (It has not been verified whether the customer base can be estimated, so I will post it somewhere again if possible)

To explain

How to get the comment attached to the video posted on the specified channel
How to get the channel registered by the specified channel

I explained it before, so don't explain it this time.

Registration part to start using API
I did it with reference to https://qiita.com/moshisora/items/4ea23d5abd7b4d852955
Get a list of videos posted on a certain channel
Specifically, how to use channels, playlistItems, videos API
- https://qiita.com/miyatsuki/items/c221b48830db2b0a9eba

Description

1. How to get the comment attached to the video posted on the specified channel

import requests
import pandas as pd

def get_comment_info(channel_id, pageToken):
    comment_url = "https://www.googleapis.com/youtube/v3/commentThreads"
    param = {
        'key': 【Browser Key】
        , 'allThreadsRelatedToChannelId': channel_id
        , 'part': 'replies, snippet'
        #If you don't need to get the reply tree, you don't need replies
        , 'maxResults': '50'
        , 'pageToken': pageToken
    }

    req = requests.get(comment_url, params=param)
    return req.json()

comment_df = pd.DataFrame([]*3, columns=["video_id", "author_name", "channel_id"])

for comment_thread in req["items"]:
    video_id = comment_thread["snippet"]["videoId"]
    
    author_name = comment_thread["snippet"]["topLevelComment"]["snippet"]["authorDisplayName"]
    author_channel = comment_thread["snippet"]["topLevelComment"]["snippet"]["authorChannelId"]["value"]
    comment_df = comment_df.append(pd.DataFrame([[video_id, author_name, author_channel]], columns=["video_id", "author_name", "channel_id"]))
    
    if "replies" in comment_thread and "comments" in comment_thread["replies"]:
        for replies in comment_thread["replies"]["comments"]:
            author_name = replies["snippet"]["authorDisplayName"]
            author_channel = replies["snippet"]["authorChannelId"]["value"]
            comment_df = comment_df.append(pd.DataFrame([[video_id, author_name, author_channel]], columns=["video_id", "author_name", "channel_id"]))

The points are the following two

In order to get the reply tree information, you need to look at snippet.replies.comments.snippet as well as snippet.topLevelComment.snippet.
However, since the above two have the same json structure, you only have to go through the same processing.

Also, in the above example, only the channel name and channel_id of the person who commented are acquired, but if you want to acquire the text of the comment, you can acquire it below.

comment_thread["snippet"]["topLevelComment"]["snippet"]["textDisplay"]
replies["snippet"]["textDisplay"]

Please check this document for details or for a slightly different usage. For example, this time we specify channel_id to get comments associated with all videos posted on that channel, but it is also possible to specify only specific videos. https://developers.google.com/youtube/v3/docs/commentThreads?hl=ja

2. How to get the channel registered by the specified channel

def get_subscription_info(channel_id, pageToken):
    subscription_url = 'https://www.googleapis.com/youtube/v3/subscriptions'
    param = {
        'key': 【Browser Key】
        , 'channelId': channel_id
        , 'part': 'snippet'
        , 'maxResults': '50'
        , 'pageToken': pageToken
    }

    req = requests.get(subscription_url, params=param)
    return req.json()

subscription_df = pd.DataFrame([]*3, 
    columns=["channel_id", "subscript_channel_name", "subscript_channel_id"])

#Passing an empty string to pageToken is the same as not specifying anything
# 1.If you are running the code you got in_id_list = comment_df["channel_id"].unique()become
for channel_id in channel_id_list:

    pageToken = ""
    while True:
        req = get_subscription_info(channel_id, pageToken)

        if "items" in req:
            for item in req["items"]:
                subscript_channel_name = item["snippet"]["title"]
                subscript_channel_id = item["snippet"]["resourceId"]["channelId"]
                subscription_df = subscription_df.append(
                    pd.DataFrame(
                        [[channel_id, subscript_channel_name, subscript_channel_id]]
                        , columns=["channel_id", "subscript_channel_name", "subscript_channel_id"]
                    )
                )


        #If the number of remaining items exceeds maxResults, nextPageToken will be returned.
        if "nextPageToken" in req:
            pageToken = req["nextPageToken"]
            print(channel_id, pageToken)
        else:
            break

Use the Subscriptions API with Japanese translation as a subcategory https://developers.google.com/youtube/v3/docs/subscriptions?hl=ja

As for how to use it, I explained how to get videoId from playlistId before, but the atmosphere is similar to that. https://qiita.com/miyatsuki/items/c221b48830db2b0a9eba#12-%E3%83%97%E3%83%AC%E3%82%A4%E3%83%AA%E3%82%B9%E3%83%88id%E3%81%8B%E3%82%89%E5%8B%95%E7%94%BB%E3%81%AEid%E3%82%92%E5%8F%96%E5%BE%97

In particular,

If you pass channelId as a parameter, you will get a list of channels that the channel is subscripted to.
However, if you receive 50 or more, you need to use pageToken. If you pass the pageToken received in the previous request, data will be returned from the continuation, so use it by concatenating on the client side
Pass an empty string at first