Keynote of “Business Analytics for Data-Driven Decision Making”

Business Analytics for Data-Driven Decision Making is one of five courses of BUx’s Digital Product Management MicroMasters® Program on edX. The course contains 7 parts released on a weekly basis, and one final proctored exam. In total, the estimated effort is more or less 30 hours.

Keynote 01: Type of analytics

Analytics is categorised into 3 following methods:

  • Descriptive analytics
    • The simplest way to define descriptive analytics is that it answers the question what has happened.
    • Descriptive analytics typically condenses large amounts of historical or real-time data into smaller, more meaningful nuggets of information.
    • The main objective of descriptive analytics is to find out the reasons behind success or failure in the past.
  • Predictive analytics
    • The next class of analytics, predictive analytics, uses data from the past to predict what will happen in the future.
    • As we can see, predictive analytics is based on a solid understanding of what happened in the past. So most organizations are deploying them after they have mastered the art and science of descriptive analytics.
  • Prescriptive analytics
    • This method is using optimization algorithms to determine the set of actions that optimize a desirable objective, given predictions of what is likely to happen.
    • As organizations gain experience and skills with data-driven business decision-making, they typically progress from descriptive, to predictive, and finally to using prescriptive analytics to inform decisions and actions in a growing number of functions.

Keynote 02: The Process of Business Analytics 

  • Define business objective
    • A process of analytics should always begin with a clearly articulated business objective.
    • When in doubt about whether we are proceeding in the right direction, we should always be referring back to this objective, and asking whether our data strategy can provide a good answer.
  • Translate business objective into a data project
    • The next phase of the process translates this business objective into a data driven decision making project.
  • Select and obtain data
    • Once we have a sound strategy planned, we need to select appropriate data. This is a crucial step that will often determine success or failure of the project.
    • For example, if our objective is to improve the response rate for a direct marketing campaign, one possible approach would be to use analytics to determine the characteristics of prospects that tend to respond to our campaign messages in the recent past, and then target our future campaigns to individuals that match those characteristics.
  • Explore data
    • After we have obtained the necessary data and decided on a methodology, we typically need to explore the data to observe some initial patterns as well as indications that perhaps some of the data contains errors that need to be corrected.
    • For example, we may find that a substantial fraction of individuals in our data do not have age information, or perhaps that a disproportionate fraction were more than 100 years old, which is likely due to input errors, or we may notice that there’s a significant gender skew that will throw off our analysis later on.
  • Transform and clean data
    • These observations need to be dealt with through a sequence of data cleaning and transformation operations.
    • For example, we may decide to throw away individuals with the wrong age, or we may simply leave their age empty and use an analysis method later on that is capable of dealing with missing data. We also need to merge the data that comes from our internal systems with any data we obtained from other sources, and bring it to the format that is required by the analysis tools.
  • Conduct data analysis
    • It is only at this point that we are ready to conduct formal data analysis. To determine what differentiates prospects who responded to our campaign from those who didn’t, we might want to try to run our data through a regression model.
    • The regression model will flag the customer attributes that have a statistically significant effect, positive or negative, on the probability that the given prospect responded to campaigns contained in our data set.
  • Transform data insights into action
    • In most cases, we will want to apply a variety of different models to look for the one that has the best predictive performance, that is the one that performs best in separating good from bad prospects.
    • Having converged on a model, we are ready to translate our analytical findings into concrete actions. Recall that our business objective is to improve the response rate of our next marketing campaign.
    • Given a list of prospective customers and their characteristics, we can use the model we developed in the previous step to assess the probability that each of those prospects will respond to our campaign, and we can limit our efforts to targeting prospects where the model’s assessment is above a threshold.
    • That way, we can save marketing resources, and if successful, we can realize superior financial results.
  • Assess outcomes
    • Last but not least, we need to assess the outcomes of our efforts.
    • Was the response rate of the campaign as high as our model predicted? If not, why not? What did we learn from the process? How can we use the learnings to be even more effective the next time around?
    • The analytics process is never a straight line, but a loop that provides the company with continuous learning and improvement.

 

Posted in IT