Statistics Online Computational Resource SOCR Capabilities Tools Services

Statistics Online Computational Resource (SOCR): Capabilities, Tools & Services Syed Husain, Alex Kalinin, Anh Truong, Matt Leventhal, Kinnothan Nelson, Selvam Palanimalai, Nicolas Christou (UCLA), Ivo D. Dinov Tools Capabilities Data Modeler Services Probability Distributome SOCR Data Dashboard SOCR Servers The main SOCR compute server, provides a unique highthroughput data processing environment (HP DL 560 Servers, 64 -bit Intel Xeon E-5 -4620 4 S/2 U (8 core), 1. 5 TB memory, 20 TB HD), 8 GB Dual Port PCI-e FC HBA, 10 Gb. E 530 FLR-SFP+ FIO NIC, Embedded Smart Array P 420 i. SOCR Modeler enables powerful data generation and dynamic model fitting using polynomial, distribution and spectral base functions. Top graph shows a Fourier model with generated data indicated by blue & fitted model indicated by red. Bottom graph illustrates parameter estimation in mixture modeling. The suite of Distributome tools (www. Distributome. org) include interfaces for (human and machine) traversal, search, and navigation of most probability distributions. It also enables distribution modeling, applications, investigation of inter-distribution relations, as well as their analytical representations and computational utilization. The entire Distributome framework is designed and implemented as an opensource, community-built, and Internet-accessible infrastructure. It is portable, extensible and compatible with HTML 5 and Web 2. 0 standards. SOCR Distributions enable the interactive visualization and computation using a wide spectrum of continuous and discrete, univariate and multivariate statistical distributions. Total: 1. 5 million Past 12 mo: 400, 000 SOCR provides a diverse suite of tools to data interrogation, protocol design and visualization, result validation & quality control. Analytical Tools Total: 30, 000 Past 12 mo: 8, 000 Learning Resources Total: 7, 500, 000 Past 12 mo: 1, 800, 000 Big Data Description URLs Research-derived, simulated, translational and clinical data archives. Dashboard for mashing multi-source socioeconomic and medical datasets, big data analytics, graphical data exploration and discovery Comprehensive collection of web-tools for demonstrating probability, statistics, mathematics and engineering concepts. These include probability calculators, statistics analysis tools, data modeling and visualization, virtual games, simulations and experiments Modern HTML 5 resources for exploratory analytics, data discovery, simulation, and visualization Community-built, open-access and multilingual resources blending information technology, scientific techniques and modern pedagogical concepts http: //wiki. socr. umich. edu/index. php/SOCR_Data http: //wiki. stat. ucla. edu/socr/index. php/SOCR_Data http: //socr. umich. edu/HTML 5/Dashboard/ http: //SOCR. umich. edu http: //SOCR. ucla. edu http: //Distributome. org http: //socr. umich. edu/Applets/index. html#Tables http: //github. com/SOCRedu http: //SOCR. googlecode. com https: //SOCRedu. atlassian. net/browse/SOCR-7 http: //socr. umich. edu/HTML 5/Brain. Viewer http: //wiki. socr. umich. edu/index. php/EBook http: //wiki. stat. ucla. edu/socr/index. php/EBook http: //wiki. socr. umich. edu/index. php/SMHS http: //socr. umich. edu/people/dinov/SMHS_Courses. html http: //wiki. socr. umich. edu http: //wiki. stat. ucla. edu/socr SOCR Consulting SOCR tools for Predictive Big Data Analytics. SOCR provide consultation on designing translational and search studies, addressing novel methodological challenges, scientific visualization, management and processing of Big Healthcare and Biomedical Data. SOCR Navigators References: http: //socr. umich. edu/html/SOCR_Publications. html Total: 80, 000 Past 12 mo: 15, 000 Computational Infrastructure SOCR Visualization Facilitate the search, discovery, traversal and utilization of over 500 separate resources including datasets, web-based tools for data modeling, calculation, experimentation, virtual games, learning modules and instructional resources. Resource Type & Usage Data & Web-services Choosing appropriate statistical tests for independent observations Nominal SOCR is supported by: University of Michigan School of Nursing, NSF grants DUE 1416953, 0716055 and 1023115, and NIH grants P 20 NR 015331, U 54 EB 020406, P 50 NS 091856, P 30 DK 089503 Predictor Variable Interactive Distribution Calculators A dozen Cloud servers provide support for all SOCR users. Over 9. 5 M (daily-unique) users have accessed SOCR resources since 2002, including 1. 7 M in the past 12 months. Categorical (>2 categories) Ordinal (Ordered categories) Quantitative Logistic Discrete regression Outcome variable Categorical Ordinal Quantitative (>2 Categories) Discrete Non-Normal Mann-Whitney or log-rank (a) Kruskal-Wallis (b) (b) (e) Spearman rank (e) Quantitative non Logistic -Normal regression (e) Quantitative Normal (e) Logistic regression Quantitative Normal Student's t test Analysis of variance (c) Spearman rank or linear regression (d) Spearman rank or linear regression (d) (e) Plot data and Pearson or Spearman rank and linear regression (e) Linear Pearson and regression (d) linear regression Statistics Online Computational Resource (www. SOCR. umich. edu), Human Behavior and Biological Sciences (HBBS), UMSN