data_8301

Recently, I worked on predicting the final driver standings for an F1 season using race results of the current season so far and historical data for the past few seasons. The linked blog post details the datasets, techniques used, and discussion on the results. Here's the github repository for the project : https://github.com/kvsingh/F1-standings-forecasting

An interesting thing I found was that the model's performance plateaued after 3 races into the season. Although this is more reflective of the model's use of the data rather than absolute predictability, it helps highlight the deterministic nature of such a technical sport.

I hope you enjoy reading it, and I would love any suggestions or comments, especially around interpretability of the results.

Thanks!

I have been job searching. A crucial component of this was comparing companies on glassdoor. But how do I compare 2 companies when one of them has a small number of ratings? Can you rely on the small number of reviews, when the other option is an established company with a lower rating but a large number of reviews?

To deal with this I used the lower bound of confidence interval around the sample mean to rank companies in Bangalore (where I live) and wrote an article about it. I hope it can be of use to the community.

Edit: The loading of library might get stuck for really large music libraries. If this is the case for you, it would be better to select the type of library with lesser songs (My saved songs vs My playlists).

Hey Hacker News!

DiscVibe emerged as a result of my indecision regarding what to listen to at any particular moment. A lot of us associate music listening with activities, and curating your own personal playlists, while fun, is cumbersome. Think of DiscVibe as your personal Spotify assistant to help you wade through the overwhelming space of familiar music and the (vast) amount of music you haven't yet listened to but would like to discover.

Spotify does have contextual mood-based playlists, but they are not personalized to your music library. The daily mix playlists and radios are personalized (and do a really good job), but they don't give the added control of context that some of us need to hit the nail on the head while taking a shower and jamming to our favorite tunes.

DiscVibe gives you a mixture of context and personalization - after logging in through your Spotify accounts, you can set your Energy Levels and your Mood (positivity) levels, (optionally) select a genre and a novelty factor (a measure of how much you are willing to explore new music vs settling on familiar tunes).

There are quite a few Spotify Playlist Creators but as far as I am aware none of them capture the "novelty" factor. It is underrated how much patience it takes to wade through a playlist of songs, none of which you have listened to before (as can be seen by my reluctance to go through any "Discover Weekly" playlists on Spotify).

As always, this is far from a polished product, and I need your feedback and criticisms to figure out which features need to be added/refined. So go ahead and Discover Your Vibe! You can also post your comments on the Product Hunt page - https://www.producthunt.com/posts/discvibe

Thanks! Same here, listened to the first two albums a lot. Had moved on to other genres by the time they released the other albums. When I tried to go back to LP, there were just too many albums to figure out which one to start with. But after doing this analysis, I feel like I should check out Hunting Party since its the closest to the first two albums.

Thanks for the tip! Will keep that in mind going forward.

That is part 2 of the blogpost. I'll be putting headphones on pandas and they'll be outputting the mood of each Linkin Park song.

Hi, author of the post here. I thought it'd be nice to post some information about the analysis.

I got some audio features (loudness, energy and valence) for Linkin Park's songs using the Spotify API. In the post, I do an album-wise analysis of these features, and how they vary across and within the albums using the Python Data Analysis Library, Pandas.

I try to articulate the general progression in style through the average change in these features, as well as the change in variation of these features.

If you're a Linkin Park fan, I would love to know how your experience of listening to LP matches with my analysis :)