Statistical Programming and Application Development
Data analysis can be done in several ways. In principle, we are programming our analyses with i.a. SAS or SPSS. Programs are typical for Analytics-as-a-Service (AaaS). Typical examples are:
- 2015: Reverse Engineering: Transformation of a low-performance, buggy and manually started non-SAS program into a high-performance, error-free and automatically triggered SAS program at a large insurance company. In view of the gravity of the situation, the SAS programming followed SOP BIO5, and its validation in accordance with SOP BIO6. To illustrate: The volume of this set of largely undocumented non-original SAS program was approximately 300 A4 pages.
- 2013: Statistical Programming: Programming of the complex extrapolation of the Census 2011. Adaptation of the cells to projected margins of 1.440 communities in 65 model variants. Visualization of a total of 93.600 models and multiples of Goodness of fit parameters. Execution as SAS Stored Process . Control via an intuitive "cockpit" with just a few "switches". Data volume:> 5,50E+09 (5,5+ billion) data lines (main application). Programming languages: SAS Macro Facility, in that SAS Base, PROC IML, SAS Hash Programming, PROC SQL and SAS procedures MEANS, TABULATE and GRAPH. Length of the two SAS programs for the main application in DIN A4 pages: 60 (ETL: 40 Analysis: 20). We are happy to provide a reference.
- Programs are the foundation of customised analyses and applications.
- Programs are transparent and facilitate documentation.
- Generally, programs are more efficient than point-and-click analyses.
- Programs allow repeated analyses at minimum extra cost.
- Client are able to integrate validated programs into their own operations and execute them there, e.g. by their own employees, automatically as a macro, as a Stored Process, or by applications such as EG, DI, or UC4.
- Optionally, programs may be expanded by other functions or integrated in special environments (Dashboards, MI or portals). Depending on the proportion of analysis (Statistics, Business) and programming, the complexity of the analysis (and know-how required), the system environment in which the data are stored, and the desired product, this role is referred to as: Statistical Analyst, Business Analyst Using SAS, Statistical Programmer, or Application Developer. The transitions are fluid.
Analysis may also include programming, e.g. as a requirement:
- Interfaces i.a. for data access (PC files, DWH tables etc.), e.g. csv/txt, ACCESS, EXCEL, Filemaker, ORACLE, SAS, SPSS, Teradata, etc.
- ETL processes (ETL: Extraction, Transformation, Load). ETL integrates distributed data for an analysis. You find examples of powerful ETL processes i.a. at Success Stories.
- Data Quality: Data undergo regular Quality Checks, i.a. including Timeliness, Consistency, Completeness, and other criteria.
- Performance: The processing is being programmed efficiently for i.a. storage, I/O and CPU (see Schendera, 2012).
Our SAS Expertise is, for example
- SAS Languages: SAS® Base, PROC SQL, SAS Macro Facility Language, Output Delivery System, PROC IML, DS2 Programming Language, SAS Annotate, Hash Programming etc.
- SAS Applications / Front-Ends: Enterprise Miner, Enterprise Guide 6.1-2, Data Integration Studio 4.3, Analyst, INSIGHT, JMP 9-5, Assist, LAB, Research Analyst, SPECTRAVIEW, Market Research Application, Time Series Forecasting System, ODS Graphics Designer/Editor, Output Delivery System (ODS) etc.
- SAS Modules: Almost all procedures for statistics, data management and visualization, above all also numerous procedures and functionalities from SAS9.4 back to SAS 6.08, for example: ACCESS®, Analyst Application, ASSIST®, Base, CONNECT®, Enterprise Guide, ETS, FSP, GRAPH, INSIGHT, IML, LAB, OR, QC, STAT.
- More applications and languages: z.B. SPSS Statistics, SPSS Modeller, KNIME, R, RWArd, R Commander, Python, HTML, UC4, Korn Shell Scripting.
For further requirements, please feel free to contact us.