Gray Matter in Data Science: Some Insights in Methodology

Capilla del Rosario, Santo Domingo Church, Quito, Ecuador

Photo by Jonathan Acuña (2018)

Gray Matter in Data Science

Some Insights in Methodology

By Prof. Jonathan Acuña-Solano, M. Ed.

School of English

Faculty of Social Sciences

Universidad Latina de Costa Rica

Saturday, August 3, 2019

Post 335 / DS Log 6

          When one talks about information found in Big Data, info can be referred as gray matter. Why gray? The answer is simply; it turns gray because in the process of making use of Data Science, when a question is asked, the response cannot be referred as black or white. In other words, the grayness is the result of having various ways of answering a question (problem) stated by stakeholders. Based on Dr. Mustaza Haider (Laureate Education Inc., 2018), a data scientist will not find answers clearly. Responses won’t be found quickly and easily; they are the gray matter that needs to be molded into the answers that institutions need to describe what is currently happening or what could come in the future.

          Based on Laureate Education Inc. (2018), Data Science Metholodology has to follow certain sequential steps to “yield” answers to questions made by stakeholders in an institution. This methodology is here to turn gray matter into a visible black-and-white “object” that can be manipulated, analyzed, and understood. Take a look at the methodological model.

1	Business Understanding	The process begins with the search for clarification about the research focus.
2	Analytic Approach	Next step is to find clarification from the stakeholders who are asking the question.
3	Data Requirements	After that, the data scientist prepares the parameters to meet the desired outcome.
4	Data Collection	The following phase consists of the actual collection of data to answer the question.
5	Data Understanding	Next stage refers to using the collected data to construct the data set for cleansing.
6	Data Preparation	The subsequent maneuver is the actual cleansing of the data from “dirty data.”
7	Modeling	Next procedure implies the development of a descriptive or predictive model for the data.
8	Evaluation	Next in line is to evaluate the depurated data to see if it provides insight into the question.
9	Deployment	Our before-last step is to “move strategically” the collected data to “push out” an answer.
10	Feedback	Finally, refinement of the model created is ready to run or to get adjusted.
Created by Prof. Jonathan Acuña with information from Laureate Education Inc. (2018)

         

          Parallel to these methodological considerations, the following infographic presents what happens when these processes (or methodological steps) are transformed into questions that need to be fully answered to get to provide an answer to what is being asked by institutional stakeholders.

          To conclude, the Data Science Methodology used to answer a question is precise and must be followed to the very detail. Doing otherwise will probably trigger answers that do not provide any light to problem a company wants to explain or an issue that can materialize in the short or long run.

References

Laureate Education Inc. (2018). Asking Questions with Data Science. Retrieved from One Faculty: https://dtl.laureate.net/webapps/blackboard/content/listContent.jsp?course_id=_165016_1&content_id=_801203_1&mode=reset

Laureate Education Inc. (2018). Things Data Science People Say. [Video File]. Retrieved from https://dtl.laureate.net/webapps/blackboard/content/listContent.jsp?course_id=_165016_1&content_id=_801203_1&mode=reset

Gray Areas in DS by Jonathan Acuña on Scribd