/keeptry-ml

Some ml demo by spark and scala.

Primary LanguageScala

keeptry-ml-demo

keeptry machine learning demos

环境

spark 2.3.1 scala 2.11 hdfs 3.1.0 环境搭建(待完善)

demos

  • Spark ALS: 基于Audioscrobbler数据集的音乐推荐
  • Spark ItemCF: 基于MoveLens数据集的电影推荐

Benchmarks

Spark ItemCF

With datasets on MovieLens 100K and 1M.

MovieLens 100K

  • RMSE = 0.9049792521615789
  • MAE = 0.7078706510784817

MovieLens 1M

  • RMSE = 0.8624739281707584
  • MAE = 0.6485303967964491

References

  1. Surprise KNNBaseline
  2. Factor in the Neighbors: Scalable and Accurate Collaborative Filtering
  3. Factorization Meets the Neighborhood: a Multifaceted Collaborative Filtering Model