FAQs -- Frequently Asked Questions

In the data science field, we work with all types of data, including relational (SQL RDBMS) and NoSQL databases and their outputs:

• Text files (CSV or fixed width),
• Spreadsheets (XLS),
• Log files,
• JSON and YAML, and
• Web files (HTML and XML).

SQL is appropriate for data that requires high reliability in transactions and is modeled using a relational standard, such as the Boyce-Codd normal form. The need for fastidious results relies on systems that provide ACID results--transactions that are Atomic, Consistent, Isolated and Durable. The financial services and medical industries remain traditional sectors that favor SQL.

NoSQL is appropriate for data that must be captured, stored and retrieved in a systematic manner but does not need ACID precision. Companies like Facebook or Twitter use NoSQL databases. Often these systems store big data--either relational or not--that must be scaled up or out while still providing reasonable accessibility and accuracy. These databases are BASE in nature--Basic Availability, Soft-state and Eventual consistency.

Questions and problems are as varied as the businesses who pose them:

• How consistent is the data?
• What are trends in database use?
• Define key indicators for customer demographics.
• What were the Q3 results?
• Is the product working within acceptable performance benchmarks?
• What are the projected outcomes?

To achieve success we often work to remove barriers through collaboration:

• Clarifying questions and problem statements with key users to make the statements SMART,
• Interviewing data owners to see if potential data sources will aid us in reaching conclusions,
• Working with database administrators and subject matter experts to identify key variables, and
• Arranging with IT operations and system administrators to access those data stores.

Every project is unique--new questions, new challenges, new answers. That's what we enjoy!

We adapt the software development life cycle (SDLC) process to address data science problems.
We might approach your project in this manner:

1. Define a project's goals and budget with your project champion.
2. Identify and refine the question or problem statement.
3. Document SMART requirements with your end users, business leaders and information systems personnel.
4. Write a project plan with timeline and deliverables.
5. Locate likely data sources within and outside your domain.
6. Design, develop, test and hone the solution in an agile, iterative cycle.
7. Deploy the solution and train your users.
8. Evaluate the project's success.

We use the R language for analytics, visualization and publication. R is an open source application created for multiple operating systems, thereby eliminating the need for clients to purchase proprietary software.

Absolutely! Melding business processes with data is our speciality.
We can create a data science process tailored to your company's needs.

One component is identifying the key players in your process. These people provide insight about data stores, logistics and end users, to name a few considerations.

• Business analyst,
• Data architect,
• Data scientist,
• Database administrator,
• Project champion,
• Stakeholder and
• System administrator.

You are the subject matter expert. You are an integral partner in the exploration and delivery process. You provide the domain knowledge to craft the questions, clarify the goals, understand the data and evalute the results.

Two words: collaboration and customization. Regardless of the data set size, type of data (SQL v NoSQL) or platform (Linux v Windows OS), our projects incorporate these attributes.

As an example, the electronic patient records management system required us to do the following:

• Collaborate with the business owner to understand the business processes.
• Define the owner's performance and quality metrics and incorporate those into the system design.
• Analyze and clean existing data.
• Develop a relational data model and n-tier SQL-based system.
• Customize user interfaces and reports with the client's corporate color scheme and logo.
• Incorporate data quality assurance and quality control measures into the system.
• Train end users to use the solution and technical personnel to maintain it.

For more project examples please read client testimonials or our capability statement.

We deliver clean data and clear communication in a reproducible format.
Deliverables can relate to many business areas:

• Executive communication
• Business workflow
• Sales and marketing collateral
• Financial analysis
• Project management
• Technical implementation
• Systems management
• User functionality and training

Several delivery methods exist for our clients in the Puget Sound or elsewhere:

• One-on-one tutoring,
• Small group settings,
• Classroom-based instruction or
• Video conferencing.

Our goal is to improve humanity's condition. We serve government, non-government (NGO), for-profit and not-for-profit organizations.

• Education
• Health care
• High technology
• Insurance
• Internet
• Life sciences
• Public policy
• Subject to U.S. Food and Drug Administration (FDA) quality assurance.

We work with non-geeks all the time. Every business unit has data.

• Sales, marketing, accounting, finance and legal :: Data analysis can help you understand the past and present and change the business. Business intelligence aggregates data and gives you key performance indicators to help you run your business. Machine learning models can predict outcomes.
• Science and product development :: Time series data analysis can shed light on the progression of research results.
• Quality assurance :: Analytics combined with process data can help build successful production systems.
• Quality control :: Analytics applied to end product performance tests can reveal areas for improving the production process.
• IS, IT and Operations :: With the rise of mobile devices and bring-your-own-device (BYOD) policies, issues of security and resource management require on-going analytics.

You have goals. We do our best to help you achieve them.

Potential benefits of combining data science with software development (SDLC) include the following:

• Increase revenues with new products and services that leverage existing data and domain knowledge,
• Reduce expenses by improving efficiency and consistency in business processes and streamlining communications between lines of business, or
• Improve quality of products and services with key benchmarks and metrics.

We explore your data and reveal knowledge through data analysis using metrics and visualization. We can predict the outcome through machine learning using statistical models that we develop following a strict methodology. We present the results and train your key persons to use the solutions.