Distributed-Matrix-Factorization

A Machine Learning project: Distributed Stochastic Gradient Descent method for Matrix Factorization in pySpark by Sudev Bohra

How to Run

• Run

$ $(SPARK) dsgd_mf.py $(NUM_FACTOR) $(NUM_WORKER) $(NUM_ITER) $(BETA) $(LAMBDA) $(TRAINV) $(OUTPUTW) $(OUTPUTH)  

Example

$ spark-submit dsgd_mf.py 20 5 100 0.9 1.0 test.csv w.csv h.csv  

Input

Takes input of matrix V in sparse format. One row, col, value per line eg. the 2x2 identity matrix would look like

0,0,1
1,1,1

Output

Outputs file w.csv and h.csv in dense matrix format