5 ms·
Ben from Kaggle. Open up the ~50 different individual datasets linked in separate tabs, and then quickly flip through all of them trying to get a sense of what
by benhamner 8y ago
Ben from Kaggle.
Open up the ~50 different individual datasets linked in separate tabs, and then quickly flip through all of them trying to get a sense of what each one is.
That experience will demonstrate one of the main challenges we're aiming to solve by making Kaggle Datasets your default place to publish data online (https://www.kaggle.com/datasets https://www.kaggle.com/datasets)
- logancg 8y agoThis is a great idea Ben, and I appreciate the work you do. Do you see Kaggle datasets as a tool to encourage better data formatting, or are you also thinking about building tools for automatically visualizing, cleaning, and organising data?
- benhamner 8y agoAll of the above, and more! One thing I'm really excited about that we're about to release is a much better explorer for tabular data (automated histograms, sorting/filtering/showing the data, and the like). We also encourage sharing analytics code and visualizations that users create on the data back to the community. For example, see all these visualizations and insights in StackOverflow's developer survey data linked from https://www.kaggle.com/stackoverflow/stack-overflow-2018-developer-survey/kernels https://www.kaggle.com/stackoverflow/stack-overflow-2018-dev...
- bhnmmhmd 8y agoI've heard that Kaggle data sets encourage people to do "supervised" ML only. Is that true?
- hideo 8y ago(Not Ben, but - ) outside of academia, the main thing that seems to encourage people to do supervised ML is that it's the only thing that seems to work. I haven't really heard of any success stories with using unsupervised techniques for most common ML applications.
- harias 8y agoWhat about clustering?
- dotancohen 8y agoI used a very simple unsupervised ML built in scikit-learn to find good matches on OK Cupid. Worked very well, it found definite boundaries between the clusters of women. One of the features was a subjective rating of how much I liked some of the women, and scikit-learn then suggested to me other women in the clusters that had my best ratings. It turns out that I like vegetarians, redheads, and left-wingers. Which happens to be true, even though I eat meat and do not identify as left-wing. But those traits correlate with _other_ traits that are more difficult to measure objectively, such as caring about children, liking to hike, and preferring an evening of sex to an evening of television.
- raverbashing 8y agoUnsupervised works, but your ability to measure "does it work or not" is much more dependent on a case by case evaluation rather than a score. (Because if you know a priori what is it that you want to measure - it's supervised)
- atupis 8y agoYeah this my experience too, evaluation ends being almost endless time sink.
- laichzeit0 8y agoI'm not an expert, but I feel that: Unsupervised techniques work really well for language modelling. There is also weakly supervised and distant-supervision, where the labels are "noisy" or not exactly what you want. You're right in that strong supervision, where you basically trust your class label, works really well, because it's probably the easiest case. Combining unsupervised (e.g. pre-trained language models) with a very small set of strongly labeled data, or a larger set of weakly labeled data, seems to work pretty well too.
- taneq 8y ago
- stuartaxelowen 8y agoNot at all - I released a customer support on Twitter dataset there specifically focused on unsupervised tasks! I think the focus on supervision in what people do with the data shows that there are still a lot of people poking around with the easier supervised tasks. [0]: https://www.kaggle.com/thoughtvector/customer-support-on-twitter https://www.kaggle.com/thoughtvector/customer-support-on-twi...
- benhamner 8y agoThe competitions we host (https://www.kaggle.com/competitions https://www.kaggle.com/competitions) are supervised and always have a target we can create a numeric leaderboard on, but the public datasets (https://www.kaggle.com/datasets https://www.kaggle.com/datasets) are used for everything under the sun. There's some supervised ML use of those, and a lot more open-ended exploration, visualization, cleaning, clustering, language modeling, etc.
- fursund 8y agoThis is great! Thanks for sharing. Would be awesome if your license filter had a "not for commercial use" vs. "for commercial use" or similar.
- mark_l_watson 8y agoGreat, thanks for the link (and to the blog author for her links). I do machine learning at work, but just two very specific use cases involving GANs and RNNs. I appreciate resources to use in my own time to explore other architectures.
- benhamner 8y agoYou can also start a new Jupyter notebook session on any of these datasets with a click (click "New Kernel"), and then accelerate your analysis by attaching a GPU to the session with another click (for applications a GPU helps, e.g. training Tensorflow models on image data)
- akarve 8y agoIronically, the same challenges are better solved in the world of code: Docker, GitHub, npm, etc. Some friends and I created Quilt to bring versioning and packaging to data: https://quiltdata.com/ https://quiltdata.com/. The interface is the familiar Python lifecycle of install and import.
- atc 8y agoI love this site. Great work. Glad you posted here as I'd probably not have discovered it as quickly.