Insightful Analysis Begins With A Right Question

Simplified flow of data analysis. Understand the requirement. Defining the needs. Analyze the information. Interpret the result. Presenting the findings. Storytelling. Recommending solutions.

How to get start to work with data? What kind of data to adopt for solving your business problem? What tool is best suit to answer your question? These questions worth a deep thought if you are serious to change your business model from a traditional way to systematic data driven strategies in this digital age.

As the data storage and management architecture and system are improving from day to day, the trend for data collection is switching from goal oriented to massive, unstructured and non-goal oriented by capturing whatever a business is being exposed to. From there onwards, subset of data is then extracted for further analysis according to the requirement. This setting is trying to avoid the loss of information. Whether or not you are currently getting ready to impose data related approach in your business, most of us believe the day where application of data is no longer a trend but a must in every business model will not be long from now. Therefore, it is vital to have awareness to capture, collect and store every data available so will not be trapped in condition lack of data for modelling.

No matter how sophisticated your infrastructures or analysis tools are, it does not bring any value to you if you failed to use it appropriately. Data is just collection of words, numbers, pictures and others, it is meaningless until you utilize it in a proper way to explore its indispensable value. Taking a scientific experiment as an example, the process is always starting with problem statement and hypothesis. Scientists are clear about what they intend to test out or to prove then they proceed with experiment and data collection. The same concept should be applied to data analysis, you are ‘blind’ until you find your ‘target’. The analysis journey can be broken down into several steps as below:

What is your problem? Who are the audience?

• Evaluate well-being of business. Have an overview of current situation.
• Who are the audience asking for the analytics job to be done? Different departments may look at each business challenge from different perspectives.
• Identify business challenges. Think broad and deep. So, you can foresee potential challenges in near future.
• Your goal corresponds to challenges. May it be business expansion and revenue boosting plans.

Asking a question?

• Previous step talks about wider business landscape, analysis should begin with a narrower, attainable and more specific focus. Properly frame your question.
• For example, instead of the goal mentioned, you may ask, what are the marketing plans to boost sales? which cost to be controlled/reduced? Which are the most profitable products?
• Focusing one target at a time, the accumulated outcome will be achieving our main goal.

Data source and data quality

• If having proper planning about data application, a business should have massive/adequate amount of data.
• Identify subset of dataset that is useful for analysis to give insights regarding the business problem.
• Data cleaning, extraction, transformation and loading to data warehouse for analysis.

Analysis modelling

• Always hold on to the principle. There is no perfect methods/models that handle all cases. Nonetheless, there is always one method works the best for each case.
• Look at the types, distribution and behavior of our data.
• Plan for proper analysis pipeline and proposed models.
• Train, test and evaluate our statistical models. Optimize and selecting the most performing one.

Result interpretation and presentation

• When doing analytics task, it is a good practice to assume our audience is not expertise in analytics.
• Presenting visualization of results for better illustration. Explaining results in plain English rather than all the technical term which may be difficult for people without statistics background.
• Ensure interpretability and expandability of findings.

Suggestion

• Outline how the results provides valuable and actionable answer to the crucial problems raised.
• Recommending solution to stakeholders.
• Documented or properly written as report for submission and record purposes.

Analytics team deal with data to aid business process along the workflows, from sales, finance, marketing to management. Data analysis is not about communicating with data solely. Talk to departments to understand their needs. It should assist them to find solution in systematic and scientific way. We do not use sixth sense, we do not guess, the results are supported by evidence.