6个数据:
点赞率:5000及格,优秀:10000点赞/小时,针对50人左右的小直播间,100人的翻倍,按这个比例算,人数递增点赞比例可以递减(大致30%)互动率:5%及格,10%优秀算法:公屏评论人数/观众总数转发率:其他的那个数据,达到5%及格,10%以上优秀关注率:观众总数小于1000人,1%及格,5%优秀观众总数大于1000人,2%及格,5-10%优秀付费率:5%及格,10%优秀算法:付费人数/观众总数留存率:整场直播停留市场(通过电脑端的抖音官网查看,每天直播都有一个平均停留时长)3分钟及格,5分钟优秀,7分钟非常优秀
6 data:
Like rate: 5,000 passing, excellent: 10,000 likes/hour, for a small live streaming room with about 50 people, double it for 100 people, and according to this ratio, the like ratio can decrease with increasing number of people (roughly 30%)Interaction rate: 5% passing, 10% excellentAlgorithm: number of comments on the public screen/total number of viewersForward rate: the other data, reaching 5% passing, more than 10% excellentFollowing rate: total number of viewers less than 1,000, 1% passing, 5% excellentTotal number of viewers greater than 1,000, 2% passing, 5-10% excellentPayment rate: 5% passing, 10% excellentAlgorithm: number of paying people/total number of viewersRetention rate: the entire live broadcast stays in the market (check through the Douyin official website on the computer, there is an average stay time for each live broadcast) 3 minutes passing, 5 minutes excellent, 7 minutes very excellent
There is no specific answer to this question, because "" depends on the data requirements and analysis purposes. If you are analyzing a simple problem or doing simple data exploration, a small amount of shallow data may be sufficient. If you are doing detailed data mining or machine learning tasks, you may need more deep data. Therefore, the amount of qualified data depends on the specific situation