Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/holgerbrandl/data_science_with_kotlin

Sources and slides of my KotlinConf2018 presentation about data-science with kotlin
https://github.com/holgerbrandl/data_science_with_kotlin

data-science data-visualization java kotlin

Last synced: 16 days ago
JSON representation

Sources and slides of my KotlinConf2018 presentation about data-science with kotlin

Awesome Lists containing this project

README 

        
# Building Data Science Workflows with Kotlin



It was a wild -- and fun!! -- ride through NYC with Kotlin at [KotlinConf](https://kotlinconf.com/), see yourself!



### >> [Slides](https://holgerbrandl.github.io/data_science_with_kotlin/data_science_with_kotlin.html) <<



### >> [YouTube Video](https://www.youtube.com/watch?v=yjVW6uCmVBA) <<



### >> [kts converted to Jupyter Notebook](report_rendering/nyc/WildRideThroughNYC.nbconvert.ipynb) <<



## Abstract



Kotlin's language design and its great tooling provide a wonderful framework for data science. Still evolving are libraries for convenient and _kotlin-esque_ table manipulation and reporting.



In this session I would like to present the design and features of `krangl`, which is a {K}otlin DSL for data w{rangl}ing. By mimicking well established concepts from pandas and R, it implements a grammar of data manipulation using a modern functional-style API. It allows to filter, transform, aggregate and reshape tabular data. Clearly static types are preferable when using Kotlin, but very often data is fluent and has no immediate type. To bridge this gap, `krangl` provides means to toggle between typed and untyped data.



As an example, we will discuss how to compete at kaggle with workflows written in Kotlin. To facilitate that, we will use Jupyter to convert Kotlin scripts into HTML/notebooks.



## Sources



For slide sources see `docs`



For example code see `src`



# Taxi challenge



We picked the taxi trip duration challengefrom Kaggle as an example



https://www.kaggle.com/c/nyc-taxi-trip-duration



## Other interesting kernels of the Taxi Competition



* [NYC Taxi EDA - Update: The fast & the curious](https://www.kaggle.com/headsortails/nyc-taxi-eda-update-the-fast-the-curious)

* [From EDA to the Top (LB 0.367)](https://www.kaggle.com/gaborfodor/from-eda-to-the-top-lb-0-367)

* [NYCT - from A to Z with XGBoost](https://www.kaggle.com/karelrv/nyct-from-a-to-z-with-xgboost-tutorial)



All NYC trip duration kernels can be found at https://www.kaggle.com/c/nyc-taxi-trip-duration/kernels.



## References



Previous Talk Pointers



*  [Kotlin's emerging data-science ecosystem](https://holgerbrandl.github.io/kotlin4ds_kotlin_night_frankfurt//emerging_kotlin_ds_ecosystem.html)  from the Kotlin Night in Frankurt (Germany) in spring 2018



Repo Pointers



* https://github.com/holgerbrandl/kscript

* https://github.com/holgerbrandl/krangl

* https://github.com/holgerbrandl/kravis

* https://github.com/ligee/kotlin-jupyter



Articles



* https://kotlinfrompython.wordpress.com/2017/05/27/why-program-in-kotlin-instead-of-python/



Figure References



* https://www.rstudio.com/



## About me



Holger Brandl works as a data scientist at the Max Planck Institute of Molecular Cell Biology and Genetics (Dresden, Germany). He holds a Ph.D. degree in machine learning, and has developed new concepts in the field of computational linguistics. More recently he has co-authored publications in high-ranking journals such as Nature and Science. He is actively contributing to the Kotlin community by developing tools and libraries for bioinformatics, high-performance computing and data-science.