Takuya Kitazawa

Better to introduce the type checker for development: http://mypy-lang.org/

- [ ] libFM - [ ] fastFM - [ ] librec

e.g., criteo click stream

Support creating User/Item object from arbitrary ID, rather than sequential index

1

FluRS should internally hold ID-index mapping.

Fix JuliaCon proceeding paper compilation error

1

https://github.com/JuliaCon/proceedings-review/issues/119

Support ranking loss for factorization machines

1

- S. Rendle, C. Freudenthaler. [Improving Pairwise Learning for Item Recommendation from Implicit Feedback](http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.587.3946&rep=rep1&type=pdf) - https://towardsdatascience.com/factorization-machines-for-item-recommendation-with-implicit-feedback-data-5655a7c749db More towards LTR. Follow up on #30

Benchmark with all {recommender, metric, dataset} pairs in the package

1

- Implicit and explicit recommendation scenario - Accuracy and running time - Comparison with third-party tools

Create a comprehensive tutorial with notebooks

1

Beyond the simple getting started sample, create an end-to-end tutorial to highlight more realistic use cases.

Something like this: ```jl struct Model data::DataAccessor hyperparameters::HyperParameters end ``` Also metrics can be parametrized in the same way, like: ```jl struct Metric k::Integer end ```

Now it's a standard library in Julia https://docs.julialang.org/en/v1/stdlib/SparseArrays/