{"id":6472,"date":"2023-12-07T10:20:59","date_gmt":"2023-12-07T10:20:59","guid":{"rendered":"https:\/\/cmotions.nl\/?p=6472"},"modified":"2025-01-09T10:48:33","modified_gmt":"2025-01-09T09:48:33","slug":"ai-voorspelt-de-top-2000-van-2023","status":"publish","type":"post","link":"https:\/\/cmotions.com\/en\/ai-predicts-top-2000-of-2023\/","title":{"rendered":"AI predicts the Top 2000 of 2023"},"content":{"rendered":"

Everyone knows it - that one great uncle or colleague who scores full points at the Christmas quiz during the music round, knows all the new songs and also effortlessly sings along with all the <\/span>golden oldies<\/span><\/i>. Would you also like to impress with your music knowledge this month? <\/span>We'll help you! How to. <\/span>It's that time of the year again!<\/span><\/i> With the year 2023 almost behind us, the cold is making its return, oliebollenkramen are springing up all over the country and ... it is once again possible to vote for Radio 2\u00b9s Top 2000.<\/span>!\u00a0<\/span>\u00a0<\/span><\/p>\n

We give away in advance that Maan, Goldband, Miley Cyrus and Nothing but Thieves will enter this year's Top 2000 very high. Curious about which songs? Then read on quickly. Use it to your advantage and impress your friends, family and colleagues with your new music knowledge and finally win that Christmas quiz this year.<\/span>!<\/span>\u00a0<\/span><\/p>\n

Trends in the Top 2000<\/h3>\n

But how do we know this? After all, everyone is free to vote for his\/her own favorite songs for the Top 2000. Here there are often external factors that influence how people vote. For example, the number<\/span> A Whiter Shade of Pale<\/span><\/i> by Procol Harum in 2021 at once from spot 153 (2020) to spot 3. This song was the favorite song of Peter R. de Vries who was shot dead that year<\/span>\u00b2<\/span>. Also, artists who have been in the news a lot in a year often provide movement within the charts.\u00a0<\/span>\u00a0<\/span><\/p>\n

With this in mind, we created a machine learning model that predicts the high entrants in the upcoming Top 2000. A model is a structured way of looking at your data to see if there are any patterns in it. We used data from the Top-40, Spotify, information about whether artists have been in the news and much more.\u00a0<\/span>\u00a0<\/span><\/p>\n

For consultants at Cmotions, creating such models is familiar fare: using Artificial Intelligence techniques, we help provide insights and make predictions based on various data sources. We do this on a daily basis in our assignments for clients, but have also done so, for example, to improve the <\/span>catch-up marriages<\/span><\/a> after Corona and to predict the new restaurants with <\/span>Michelin Stars<\/span><\/a> designate.\u00a0<\/span>\u00a0<\/span><\/p>\n

Predicted high new entrants<\/span><\/b>\u00a0<\/span><\/h3>\n

The big question, of course, is: which songs do we think are the high new entrants?<\/span> Based on historical patterns, we predict that the songs below are high new entrants into this year's Top 2000! And Stiekem (pun intended) we are quite pleased with our model's predictions. Indeed, Flowers (worldwide) and Stiekem (in the Netherlands) were the most streamed songs on Spotify in 2023. Hopefully the results will show that we are right!<\/span>\u00a0<\/span><\/p>\n

Predicted high new entrants Top 2000<\/span><\/b>\u00a0<\/span><\/p>\n\n\n\n\n\n\n\n\n\n\n
Artist<\/span><\/b>\u00a0<\/span><\/td>\nNumber<\/span><\/b>\u00a0<\/span><\/td>\n<\/tr>\n
Moon & Goldband<\/span>\u00a0<\/span><\/td>\nSecretly<\/span>\u00a0<\/span><\/td>\n<\/tr>\n
Felix Jaehn & Ray Dalton<\/span>\u00a0<\/span><\/td>\nCall it Love<\/span>\u00a0<\/span><\/td>\n<\/tr>\n
Meteor & Hannah Mae<\/span>\u00a0<\/span><\/td>\nWhat do you want from me?<\/span>\u00a0<\/span><\/td>\n<\/tr>\n
Flemming<\/span>\u00a0<\/span><\/td>\nParacetamoles<\/span>\u00a0<\/span><\/td>\n<\/tr>\n
Son Mieux<\/span>\u00a0<\/span><\/td>\nTonight<\/span>\u00a0<\/span><\/td>\n<\/tr>\n
Nothing but Thieves<\/span>\u00a0<\/span><\/td>\nOvercome<\/span>\u00a0<\/span><\/td>\n<\/tr>\n
Miley Cyrus\u00a0<\/span>\u00a0<\/span><\/td>\nFlowers<\/span>\u00a0<\/span><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Predictions? Our approach<\/span><\/b>\u00a0<\/span><\/h3>\n

To make a prediction based on patterns, you need data, preferably a lot of and relevant data. So we started collecting data. First, we collected the lists of the Top 2000 from the years 2013 to 2022. For each year we looked at the position of all the songs and their corresponding artist(s). In addition, we collected data for the same period from the weekly Dutch Top 40, based on the idea that new or popular songs enter the in the Top 40 first and then people vote for them for the Top 2000. We also extracted additional information about the collected Top 2000 and Top 40 songs from Spotify. These include the style of a song, such as \"danceability\" and to what extent a song is instrumental. Finally, we collected data on deceased artists and on the amount of Google searches per year about the artists in our dataset.\u00a0<\/span>\u00a0<\/span><\/p>\n

From our own experience, we know that most of the time and value is in properly preparing the data before we can create a model. For example, one challenge in this process was that names of songs and artists are often written in different ways. This happens when using various sources, but even within the Top 2000 the spelling is not consistent. For example, the song Knocking on Heaven's Door by Guns N' Roses was still written as Knockin' on Heaven's Door through 2020. And \"Under Pressure,\" is that by Queen & David Bowie or by David Bowie & Queen? We came across both.<\/span>\u00a0<\/span><\/p>\n

With this complete data set, we created a model based on the data from the years 2013 through 2021, and then had the model make predictions about the year 2022 to validate whether our predictions were correct. That already looked promising! Finally, we made predictions for this year's edition in the same way.<\/span>\u00a0<\/span><\/p>\n

Predictive value variables<\/span><\/b>\u00a0<\/span><\/h3>\n

In total, we used no less than 171 variables for our predictions. Using the right AI techniques, we arrived at our predictions for this year \u00e1nd we gained insight into what determines whether a song can make a big hit in the Top 2000. For those interested: at a later date we will share more details about the techniques we used in preparing the data, which machine learning techniques we considered, and how we arrived at insights with SHAP about what determines whether a song will rise to a high ranking upon entry. In this article, we'll stick to some interesting insights.<\/span>\u00a0<\/span><\/p>\n

What is striking in the graph below is that different types of characteristics recur in the list of best predictors according to the model. <\/span>Spotify audio features,<\/span><\/a> such as <\/span>Duration<\/span><\/i>, <\/span>Loudness<\/span><\/i>, <\/span>Energy<\/span><\/i>, <\/span>Acousticness <\/span><\/i>and <\/span>Pace<\/span><\/i>, describe the songs very technically and therefore do matter to get (high) in the Top 2000! In addition, the number of weeks a song has been in the Top 40 in recent months is very predictive of where it will enter the Top 2000. But also to what extent there <\/span>'buzz' <\/span><\/i>around an artist is - measured by us with the <\/span>Google Trends<\/span><\/a> dates - proved predictive of the debut listing.<\/span>\u00a0<\/span><\/p>\n

\"\"<\/p>\n

With the techniques we used, we discovered not only which songs will rank highly and which characteristics are most relevant, but also how those characteristics affect the prediction. For example, the graphs below show the relationship between the increase in listing, on the one hand, the number of weeks in the Top 40 over the past year (left) and, on the other hand, the Loudness <\/em>of a song (middle). From this we learn that if a song has been in the Top 40 for more than six months in the past year, its chances of coming in high in the Top 2000 are greatly increased. And the Loudness <\/em>and Acousticness <\/em>show that quieter songs are more likely to receive high ratings than busy, energetic songs.<\/p>\n

\"\"\"\"\"\"<\/span><\/span><\/p>\n

Full of expectation, our hearts beat<\/strong><\/h3>\n

At the time of writing, voting is still open for the 2023 Top 2000, so whether we're actually going to be right we'll have to wait and see. Hopefully the above did inspire you, and should Secretly from Moon and Goldband<\/strong> indeed come in high then you already know where to find us next year!<\/p>\n

Are you curious about more details on exactly how we went about this and the underlying data and prediction model? Then keep an eye on our posts!<\/p>\n

 <\/p>\n

\u00b9<\/span><\/span>\u00a0<\/span><\/span>Due to the 25th anniversary of this list of lists, this year's Top 2000 actually consists of 2,500 songs<\/span><\/span>\u00a0<\/span><\/p>\n

\u00b2 https:\/\/www.nporadio2.nl\/nieuws\/top2000\/7d99ad03-980f-4321-8d0c-580bc643699b\/peter-r-de-vries-geeerd-in-top-2000-met-a-whiter-shade-of-pale\u00a0<\/a><\/p>","protected":false},"excerpt":{"rendered":"

Everyone knows them - the great-uncle or colleague who scores full points at the Christmas quiz during the music round, knows all the new songs and also effortlessly sings along with all the golden oldies. Do you also want to impress with your music knowledge this month? We will help you! How to. It's that time of the year [...]<\/p>","protected":false},"author":2,"featured_media":9810,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_et_pb_use_builder":"","_et_pb_old_content":"","_et_gb_content_width":"","footnotes":""},"categories":[51,55],"tags":[],"class_list":["post-6472","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-artikel","category-data-science-ai-nl"],"_links":{"self":[{"href":"https:\/\/cmotions.com\/en\/wp-json\/wp\/v2\/posts\/6472","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/cmotions.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/cmotions.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/cmotions.com\/en\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/cmotions.com\/en\/wp-json\/wp\/v2\/comments?post=6472"}],"version-history":[{"count":2,"href":"https:\/\/cmotions.com\/en\/wp-json\/wp\/v2\/posts\/6472\/revisions"}],"predecessor-version":[{"id":9641,"href":"https:\/\/cmotions.com\/en\/wp-json\/wp\/v2\/posts\/6472\/revisions\/9641"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/cmotions.com\/en\/wp-json\/wp\/v2\/media\/9810"}],"wp:attachment":[{"href":"https:\/\/cmotions.com\/en\/wp-json\/wp\/v2\/media?parent=6472"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/cmotions.com\/en\/wp-json\/wp\/v2\/categories?post=6472"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/cmotions.com\/en\/wp-json\/wp\/v2\/tags?post=6472"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}