Research Analysis

'Research and development is a crucial task when designing data science outcomes.'

 Statistical design theory was a generic way to emphasize the importance of data driven decision making. Too often, data is collected with no regard as to how the data should be analyzed, rather than could be analyzed. The data sovereignty framework key indicator Data Management encompasses the idea of design of experiments, data collection and practice, as well as establishing ways of continually refining and testing collected data to optimize explanatory power and minimizing error variance.

This was what is meant by statistical design theory: using current and established statistical principles to guide all types of data analysis. Data science is the act of incorporating these principles to exact an outcome. The benefits of this approach will undoubtedly provide more meaningful insights to data driven decision making.

Since there are many stages of data collection or practice that are always ongoing in many organizations; I think it is important to consider strategic management and planning when incorporating data science into workflows. The brief workflow below is an example how I would approach an initial consultation with any client needing the expertise of a data scientist.

To evaluate the strengths and weaknesses of any statistical process, it is imperative to develop a statistical methodology framework which usually follows a general hierarchy of:

1. Examining any descriptive statistics pertinent to a particular study
2. Develop and test a number of Exploratory Data Analysis (EDA) techniques that provide a basis for more complex methodology and inference
3. Using the information from the EDA, to create a formal inferential hypothesis for examining more complex processes that may exist beyond the first stage of the current project.
4. Repeat the process until a strategic assessment has been developed
5. Design a strategic plan to implement a designed data domain

In conclusion, the goal is to move beyond simple descriptive measures and begin a more robust process that favors data driven decision making using citizen science. Mato Ohitika Analytics can provide a number of researched methodologies in statistics from basics of hypothesis testing to more advanced topics of modeling such as logistic regression, geospatial modeling, etc.