muxi166 / sparseml

Spark MLlib code optimized to efficiently support sparse data

SparseML

Yuhao Yang ([email protected])

From purchase history to movie ratings, data sparsity has always been one of the primary characteristics of big data. Powerful as Spark is on parallel processing for the partitioned data, many of the algorithms in MLlib are implemented based on the assumption of certain degree of data density, such like the gradients of logistic regression, or cluster centers of KMeans. Yet during collaboration with some Spark users, we often find their feature number at the dimension of millions or even billions, which far exceeds the capacity of some important algorithms in MLlib, or become impractical due to enormous memory consumption even with great sparsity in the training data. To fill the gap, we present a Spark package containing some major improvements we have conducted to support the sparse data at large scope. Through optimization on data structure, network communication and arithmetic operation, we can extensively compress the memory consumption and reduce computation cost for sparse data, thus to enable the algorithms on larger feature dimensions and scope. Two of the examples are the successful support of our implementation on logistic regression with 1 billion features and KMeans with 10M features and hundreds of clusters. We’ll also share some work we are contributing to Spark and some best practices we have accumulated in the context of sparse data support on Spark MLlib.

Usage:

The class/function signature remains the same as in Spark MLlib. Please refer to the examples folder

Performance:

Although the concrete performance improvements depends on the sparsity of the dataset. The algorithms in SparseSpark generally significantly reduce the time and memory consumption compared with the original Spark implementation.

Accuracy

The optimization does not affect the accuracy. It yields the same result with the Spark version, yet with less computation resources.