Analytics Spectrum in Data Science


In data science, analytics is a key discipline that explore, measure and deliver value from data, which could be in the form of structured, unstructured and semi-structured data. It is a science of investigating raw data and turn it into valuable insight for decision-making, optimization and forecasting. There are four main spectrum of analytics in the world of data science:

  • Descriptive

  • Diagnostic

  • Predictive

  • Prescriptive


Descriptive Analytics

Descriptive Analytics is a branch of statistics that summarizing or describing the data. It is a process of interpreting the historical data to obtain insights, and deliver information of what currently happened in the business. Descriptive analytics is one of the most basic but important piece of business intelligence that the company should have. There are two common measure use in statistical analysis, which are a) Measures of Central Tendency; and b) Measures of Variability.


  • Measures of Central Tendency - Mean, Median, Mode

  • Measures of Variability - Absolute Deviation from Mean, Variance, Standard Deviation, Quartile, Skewness.


Example of the question that can be raise for descriptive analytics:

  • What is the total sales and revenue for Item A from January to December this year?

  • What is the average age that subscribe to our YouTube channel?

  • Which branch of our store that generate higher profit for this year?


Example of statement in descriptive analytics:

  • The ROI for this year is USD10M, which is 3% increment compared to previous year achievement.

  • The range of age that watch for our published advertising for more than 15 second is between 23 to 28 years old.


Diagnostic Analytics

Diagnostic analytics is an advance form of analytics that evaluate information as a response to the key question that asking why things are happening. It is an exploratory phase to find the possible causes of events occurred and result of behaviours, by drill-down into deep source of information, including the external factors.


Example of question at the stage of diagnostic analytics:

  • What is/are the possible reason(s) of drastic drop of Item B in the month of June every year?

  • Why there is a sudden spike of operational cost, about 8% higher in the fourth quarter of this year?

  • What might be the cause of customer churn of using the credit card service in Bank A?


Example of statement in diagnostic analytics:

  • Item B is a seasonal product and fully functional to be utilised during cold season.

  • At the fourth quarter of this year, the workload for the production increase due to increasing of demand by the customer. Thus, it clearly a strong reason why the cost of operation increase at 8%.


Predictive Analytics

Predictive Analytics is a part of advance analytics spectrum which is apply to make predictions of the outcome in the future events. Currently, machine learning is the powerful technique used in the field of data science, and it is reliable on the historical dataset. These historical dataset will be trained and the learning model will be used to predict output or response for the new incoming dataset.


Example of questions in predictive analytics:

  • What will be the market value of the residential properties in City A for the next 5 years in 2025?

  • Who will win the presidential seat for the next coming term in US?

  • Will increasing the amount of investment in marketing budget by 20% can increase our profit gain by 10%?


Example of statements in predictive analytics:

  • For the next five years, the properties price in the City A will be increase at average rate of 5% per year.


Prescriptive Analytics

Prescriptive Analytics is a process of evaluating the information or insight and suggesting actionable recommendations on how to optimize business practices to suit multiple predicted outcomes. The process is combining the method of machine learning, business qualitative and quantitative parameters, and business intelligence to simulate actionable approaches for the possible results. A strong understanding on the descriptive statistics and prediction model are important and critical factor that significantly affect the quality of business decision.


Example of business query in prescriptive analytics:

  • What should be done in order to win the presidential seat election? Which approach that will have higher impact to the result?

  • What we can do to maximize the profit of our company this year?

Example of business statement in prescriptive analytics:

  • In order to maximize the profit of our business, the marketing budget should be increased into 10% for Item A. The highly recommended channel of marketing for this product is via mass media communication.

0 views0 comments