Factorized Point Process Intensities: A Spatial Analysis of Professional Basketball

Authors: Andrew Miller, Luke Bornn, Ryan Adams, Kirk Goldsberry

ICML 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our data consist of made and missed field goal attempt locations from roughly half of the games in the 2012-2013 NBA regular season. These data were collected by optical sensors as part of a program to introduce spatio-temporal information to basketball analytics. We compare the fit of the low rank NMF reconstructions and the original LGCPs on held out test data in Figure 5.
Researcher Affiliation Academia Andrew Miller ACM@SEAS.HARVARD.EDU School of Engineering and Applied Sciences, Harvard University, Cambridge, USA Luke Bornn BORNN@STAT.HARVARD.EDU Department of Statistics, Harvard University, Cambridge, USA Ryan Adams RPA@SEAS.HARVARD.EDU School of Engineering and Applied Sciences, Harvard University, Cambridge, USA Kirk Goldsberry KGOLDSBERRY@FAS.HARVARD.EDU Center for Geographic Analysis, Harvard University, Cambridge, USA
Pseudocode No Given point process realizations for each of N players, x1, . . . , x N, our procedure is 1. Construct the count matrix Xn,v = # shots by player n in tile v on a discretized court. 2. Fit an intensity surface λn = (λn,1, . . . , λn,V )T for each player n over the discretized court (LGCP). 3. Construct the data matrix Λ = ( λ1, . . . , λN)T , where λn has been normalized s.t. P λn = 1 4. Find B, W for some K such that WB Λ, constraining all matrices to be non-negative (NMF).
Open Source Code No To compare various NMF optimization procedures, the authors used the r package NMF (Gaujoux & Seoighe, 2010).
Open Datasets No Our data consist of made and missed field goal attempt locations from roughly half of the games in the 2012-2013 NBA regular season. These data were collected by optical sensors as part of a program to introduce spatio-temporal information to basketball analytics.
Dataset Splits No We compare the fit of the low rank NMF reconstructions and the original LGCPs on held out test data in Figure 5. For each fold, we held out 10% of each player s shots, fit independent LGCPs and ran NMF (using the KL-based loss function) for varying K.
Hardware Specification No No specific hardware details (GPU/CPU models, memory amounts) used for running experiments are mentioned in the paper.
Software Dependencies Yes To compare various NMF optimization procedures, the authors used the r package NMF (Gaujoux & Seoighe, 2010).
Experiment Setup No Though we have described a continuous model for conceptual simplicity, we discretize the court into V one-squarefoot tiles to gain computational tractability in fitting the LGCP surfaces. To overcome the high correlation induced by the court s spatial structure, we employ elliptical slice sampling (Murray et al., 2010) to approximate the posterior of λn for each player, and subsequently store the posterior mean.