Just heard the ad for the chart today on radio 1 and not like us isn’t even mentioned, it’s between Lola and Gaga as per the ad which seems odd

 

I can't allow this to give me false hope :teresa:

They probably they make it after Wednesdays update which had Kendrick miles away but we shall see

I don't think having a weekly 300,000 lead on Spotify is enough for Kendrick

True at some point had > 1 million lead and still lost

 

I kinda suspect lots of free streams for Kendrick

 

 

I’ll be very surprised if Kendrick is below Gaga, but I’ll also be surprised if he’s #1, and I definitely don’t have any faith in the Thursday estimation machine doing “Not Like Us” justice.

In general I think they just take an average for the 6 days than actually estimate what Thursday would be

They’ve said that they don’t and take into account trends during the week, but we’ve no idea what that really means as they haven’t given any details.

There’s no way Kendrick is legitimately behind GaGa, the question is surely whether he is #1 or #2?

 

Gaga’s streams have dropped significantly below Kendrick’s over the full week. Even with an average of 6 days that will place him higher again for Thursday; I imagine they have just mentioned GaGa as it’s a new song, to be within contention.

Edited February 14, 2025Feb 14 by BellyClarkson

it's machine learning that predicts Thu streams

I've heard it explained twice, once at a data science conference in London and then at that Measure of Music thing

which is data analytics of music specifically

 

for instance, silly example

take the Titanic. First you feed data so the algorithm can learn, so you enter all the data of everybody on the Titanic

age, sex, male, female, social status, rich poor etc and if they survived or not

so the machine learns

and then you enter your own data, Im 40, guy, poor, and you ask, would I have survived the Titanic+

then the machine uses what it has learns and tells you, no, sorry you will die

(cos the machine learned that all poor guys died for instance)

 

actually it's super easy to do

definitely not an average of the week

You keep saying this but just saying 'it's machine learning' does not actually explain what exactly they do to come up with the numbers. I really don't think it's particularly sophisticated.

it's machine learning that predicts Thu streams

I've heard it explained twice, once at a data science conference in London and then at that Measure of Music thing

which is data analytics of music specifically

 

for instance, silly example

take the Titanic. First you feed data so the algorithm can learn, so you enter all the data of everybody on the Titanic

age, sex, male, female, social status, rich poor etc and if they survived or not

so the machine learns

and then you enter your own data, Im 40, guy, poor, and you ask, would I have survived the Titanic+

then the machine uses what it has learns and tells you, no, sorry you will die

(cos the machine learned that all poor guys died for instance)

 

actually it's super easy to do

definitely not an average of the week

 

All that for an estimate of one day streams? Highly doubtful

 

Yeah “machine learning” sounds swish but presumably if the inputs are basic so is the output. There have been a few instances that have made me doubt the estimates are very sophisticated. The one that always springs to mind is when “Three Lions” had done incredibly well during the week but by Thursday England had been knocked out and its streams had all but dried up, but it still seemed to gain thousands of chart sales at the end of the week.

All we have ever *officially* heard is that it's based on trends over the week. Nothing about machine learning or anything like that - that's simply assumptions.

I am assuming it looks at the percentage change Monday to Wednesday, applies that, and adjusts for the average streaming difference on Thursday (eg if Thursday is usually a quiet day or busy day).

What time do Radio 1 get sent the chart on Friday?

 

They could be basing that on the last midweek.

What time do Radio 1 get sent the chart on Friday?

 

They could be basing that on the last midweek.

 

Apparently they get the charts at around 10:30. However that was in 2022 so that could have changed since

You keep saying this but just saying 'it's machine learning' does not actually explain what exactly they do to come up with the numbers. I really don't think it's particularly sophisticated.

 

it's like the Titanic

Mr A was 51, male, rich, in category A, survived

Mr B was 18, male poor category C, died

you do the same for all people so the algorithm learns

then you ask, Im 22 male poo, would I die? yes

 

so basically the algorithm needs data to learn

for music is the same

if you give them all the data ever,

like song 1 in week 1 it did Friday 400K Saturday 325K Sunday 200K Mon 350K Tue 320K Wed 300K Thu 300K

then you do it for all songs all weeks

then the machine can learn to predict what will be the Thu sales based on the pattern from Fri-Wed

 

once a year theres this conference called Measure of Music, https://conference.measureofmusic.com/

it's for people that wanna work as data analyst in the music industry

or have an idea of an app or something but always music-linked

you go and ask, they always explain it

 

 

I'd guess Kendrick will get to #2, pushing Gaga and the new entries down but there will be no new chart topper. The least interesting outcome always seems to happen.