8 minutes
2024-02-03 (Last updated: 2024-02-04)
A Follow Up on my Hottest 100 Predictions
In my last post I ended by planting my flag and making my predictions for the Hottest 100 for 2023. And on first glance, I’m pretty happy with myself, picking not only the top song, but a good chunk of the top 20. So today I just want to do a quick follow up on how I did.
Top 20
## Song Artist Predicted
## 1 paint the town red doja cat 1
## 2 the worst person alive g flip 55
## 3 saving up dom dolla NA
## 4 rhyme dust mk 72
## 5 prada cass 90
## 6 adore u fred again 40
## 7 what was i made for? billie eilish 3
## 8 rush troye sivan 10
## 9 lovin on me jack harlow 17
## 10 chemical post malone 14
## 11 vampire olivia rodrigo 2
## 12 rumble skrillex NA
## 13 kill bill sza 9
## 14 atmosphere fisher NA
## 15 nanana peggy gou 5
## 16 sprinter dave 4
## 17 back on 74 jungle NA
## 18 eat your man dom dolla NA
## 19 therapy budjerah NA
## 20 sorry instead spacey jane NAThe top 10 seems okay, but i did miss a lot of the top 20 completely. I seemed to be undervaluing Australian artists such as Dom Dolla, Spacey Jane, and G Flip, as well as EDM as a genre, which made up a much greater portion of the top 20 than I predicted.
Snubbed Songs
## Song Artist Predicted
## 1 love type poolclvb 6
## 2 super ego babe rainbow 8
## 3 adored royel otis 15
## 4 lost without you san cisco 21
## 5 super-vision dice 22
## 6 blak britney miss kaninna 24
## 7 ruthless hooligan hefs 26
## 8 attention doja cat 27
## 9 james dean tash sultana 28
## 10 do it again benee 33
## 11 i wish you roses kali uchis 35
## 12 dash of speed rum jungle 36
## 13 too much the kid laroi 37
## 14 eyes ahead dice 38
## 15 pets and drugs the rubens 42
## 16 into your room holly humberstone 44
## 17 up lee 46
## 18 pegasus arlo parks 47
## 19 exploding angie mcmahon 48
## 20 calling metro boomin 50
## 21 must be nice ruel 51
## 22 better love eliza rose 52
## 23 your funeral maya 53
## 24 don't let me down gus dapperton 54
## 25 conceited lola young 56
## 26 messed up holy holy 57
## 27 lost in the rush telenova 58
## 28 candle flame jungle 59
## 29 prescription remi wolf 63
## 30 change laurel 64
## 31 angel pinkpantheress 65
## 32 sinner the last dinner party 66
## 33 midwest vacations 67
## 34 pretty girl ice spice 68
## 35 i wanna dance royel otis 71
## 36 make up your mind cordae 74
## 37 viper room thornhill 75
## 38 weightless arlo parks 76
## 39 can't play myself skepta 78
## 40 black mascara. raye 79
## 41 feel alive cat 80
## 42 tied up! genesis owusu 81
## 43 mrs. hollywood gojo 84
## 44 am i dreaming metro boomin 85
## 45 comma's onefour 86
## 46 lola maya 87
## 47 it's cool to be in love greta stanley 88
## 48 since i have a lover 6lack 89
## 49 lil boo thang paul russell 91
## 50 going kokomo royel otis 93
## 51 all my life lil durk 94
## 52 brain freeze northeast party house 95
## 53 healing molly millington 96
## 54 sticky kito 97
## 55 rakata the jungle giants 98So there were 55 songs in my predictions that didn’t make it into the countdown, including 3 of my top 20. Funily enough though, a lot of my predictions seemed to line up with peoples opions online with Love Type, Super Ego and Adored all being mentioned as snubs from the hottest 100.
Surprise Songs
## Song Artist rank
## 1 saving up dom dolla 3
## 2 rumble skrillex 12
## 3 atmosphere fisher 14
## 4 back on 74 jungle 17
## 5 eat your man dom dolla 18
## 6 therapy budjerah 19
## 7 sorry instead spacey jane 20
## 8 be your man g flip 22
## 9 take it off fisher 23
## 10 rich baby daddy drake 25
## 11 rough g flip 26
## 12 dance the night dua lipa 28
## 13 say yes to heaven lana del rey 29
## 14 not strong enough boygenius 30
## 15 get him back! olivia rodrigo 34
## 16 baby again.. fred again 35
## 17 boy's a liar pt. 2 pinkpantheress 36
## 18 laced up hilltop hoods 37
## 19 scary movies the rions 40
## 20 pedestal lime cordiale 42
## 21 popular the weeknd 43
## 22 sweetheart old mervs 47
## 23 padam padam kylie minogue 48
## 24 australia g flip 50
## 25 ten fred again 51
## 26 i used to be fun teen jesus 52
## 27 lookin' out king stingray 53
## 28 more than you know blink182 54
## 29 all-american bitch olivia rodrigo 56
## 30 darkside bring me the horizon 57
## 31 lost bring me the horizon 58
## 32 a&w lana del rey 60
## 33 fall at your feet peking duk 64
## 34 real life g flip 65
## 35 leaving the light genesis owusu 69
## 36 snooze sza 70
## 37 speedracer teenage dads 74
## 38 nobody gets me sza 75
## 39 sofa king royel otis 76
## 40 i don't wanna be ... ruel 77
## 41 bleed the kid laroi 78
## 42 video killed the ... teenage dads 79
## 43 7 days g flip 80
## 44 like a girl does peach prc 81
## 45 exes tate mcrae 82
## 46 the summoning sleep token 83
## 47 midnight driving teenage dads 88
## 48 nightmare polaris 90
## 49 did you know that... lana del rey 91
## 50 strawberry daydream pacific avenue 92
## 51 no bad days the terrys 93
## 52 welcome to the dcc nothing but thieves 95
## 53 stay blessed genesis owusu 97
## 54 cool about it boygenius 98
## 55 i miss you slowly slowly 99With 55 snubs, we are going to have 55 surprise songs. There doesn’t seem to be a massive trend here. G flip only made it into my countdown twice, so 5 of their songs are in this list. Its also interesting seeing which conventionally popular songs are part of this list. Boy’s a liar pt. 2 by Pinkpantheress and All American-Bitch which peaked at 2 and 10 on the aria charts were left out in my predictions, even though it predicted similar chart toppers in the top 10.
Looking at these plots, it seems that the chart data for these songs didn’t correctly join with the play data from triple J. After cleaning the names, I was hoping that there wouldn’t be too much of a discrepancy. However, in the ARIA charts, “all-american bitch” is listed as “all-american b**ch,” and “boys a liar pt. 2” is listed as “boys a liar”. This kind of discrepancy is probably present throughout my dataset and may have led to some major inaccuracies. However, it is also just part of life when dealing with text data.
Did I do better than Warm Tuna?
Part of my mission when setting out to make these predictions was to outperform 100 Warm Tunas, who utilize a compilation of social media posts to formulate their predictions.
## Song Artist rank warm_tuna my_rankning
## 1 paint the town red doja cat 1 9 1
## 2 the worst person alive g flip 2 11 55
## 3 saving up dom dolla 3 18 NA
## 4 rhyme dust mk 4 5 72
## 5 prada cass 5 19 90
## 6 adore u fred again 6 6 40
## 7 what was i made for? billie eilish 7 4 3
## 8 rush troye sivan 8 1 10
## 9 lovin on me jack harlow 9 49 17
## 10 chemical post malone 10 28 14
## 11 vampire olivia rodrigo 11 14 2
## 12 rumble skrillex 12 3 NA
## 13 kill bill sza 13 20 9
## 14 atmosphere fisher 14 23 NA
## 15 nanana peggy gou 15 15 5
## 16 sprinter dave 16 22 4
## 17 back on 74 jungle 17 13 NA
## 18 eat your man dom dolla 18 59 NA
## 19 therapy budjerah 19 8 NA
## 20 sorry instead spacey jane 20 21 NAStraight away, I can see that warm tuna did better than me, but by how much?
So, I made up a quick statistic to see how far off our predictions were. This is the sum of the magnitudes of the differences between the predicted score and the actual score. If a song didn’t make the top 100, it’s given the equivalent rank of 101. I then divide this by 100 to get the average deviation for each prediction.
## my score: 36.28## warm tuna's score: 26.5So from these statistics, we can see that my predictions were, on average, about 10 places more off than warm tuna’s.
Next Year?
I reckon this method still has promise, but I need to sort out the name joining issue to ensure that my method is working at its maximum potential. I also want to include genre and artist country into it since it seemed to be an important factor in the final rank that I didn’t account for.