3 ms·
(MLDB product lead here) I hadn't seen RecDB, thanks for the link! At a glance, other than the licensing, the main point of comparison is that MLDB is more pow
by nicolaskruchten 11y ago
(MLDB product lead here)
I hadn't seen RecDB, thanks for the link! At a glance, other than the licensing, the main point of comparison is that MLDB is more powerful/general-purpose than RecDB, as it provides primitives for not only for creating recommenders but to do other machine-learning tasks like classification, clustering, dimensionality reduction, visualization etc. MLDB also seems to have a bit more documentation :)
- boomzilla 11y agoI've used madlib[1] before and was quite happy with that. Maybe you can do a feature comparison with similar existing solutions as many people don't have the bandwidth to go through your documentations. [1] http://madlib.net/ http://madlib.net/
- nicolaskruchten 11y agoThat's a good idea, thanks! What kinds of problems were you solving with madlib?
- boomzilla 11y agoClassification problems in a medical domain. It literally took me 30 minutes to install madlib and run logistic regression on a table. It then probably took me a few more days to do some analysis, try out various algorithms and productionize the predictor.
- philgo20 11y agoA feature comparison with other solutions is a great idea. We'll work on that and share here.