In God we trust. All others must bring data – W. Edwards Deming

We just celebrated the National statistics day on the 29th June, 2016. India’s vice president Mr. Hamid Ansari, spoke about the declining quality of research; particularly in the social sciences and this was attributed to the collection, collation and computing of data- in recently concluded International seminar on Social Sciences research, conducted by the Indian statistical Institute(ISI).  In the yesteryears, most decision  were taken  intuitively , outcome drawn based upon intuition and experience. The accuracy suffered enormously due to bias of the manager. It’s needless to mention about the accuracy of such results.  Innovations in computing and advances in software, especially for statistical computing like R, SPSS, SAS etc, have brought in tremendous reliability and predictability in the field of management. In a world arguably characterized by VUCA {Volatility-Uncertainty-Complexity-Ambiguity} it is important that problems of research must be resolved in a scientific and systematic basis. Fortunately we have to our rescue some advanced techniques in statistics and machine learning for analysis of data.

Good data is extremely important to arrive at accurate decisions. Let’s take a quick peep into what exactly is this “data to decision making” concept.

Data to Decision Making

The entire process can be trifurcated into three stages. At the first stage called “Exploring the data”: we collect raw data from the field according to our objectives and variable requirements. Choice of right sample size and appropriate respondents must be given due attention. Next step is to perform the exploratory data analysis(EDA).   By figuring out whether the research is based on single variable or two variables, we need to spread them either as “Charts” or “Tables”, which arrives patterns, This essentially leads us to the second stage namely “Confirming the pattern”. The stage is a bit critical for it involves data purging and preparation.  Check for data normality and outliers are taken care of at this point to helps us discover a pattern around the data.

We are now ready for testing our hypothesis, which can be visualized as either a relationship or difference between two variables of our study.   Refer the table below,  for choice of techniques

Choice of Bivariate Techniques

Base Stats 2 Groups Greater than

2 groups

No group
Mode or Count Z proportion Chi square na
Mean T test Anova Correlation
Median / Rank Mann whitney Kruskal wallis Rank correlation

Final stage is “model development and deployment”. In the third stage, the focus is on multivariate, which can be classified into three based on our approach as

Analysis based upon the aforementioned approach leads us to “Knowledge discovery”. It provides deeper insights about data, empowering the researcher to read underlying factors and understand causality. Conclusively, the data thus obtained and prepared will be fed into the tools to envisage a model, which can be tested and deployed for assured outcomes.   To put it in simple diagram for quick grasp,  refer the following diagram.  This kind of systematic approach reduce the human bias while taking decisions in business scenario.

Types of Analysis Diagram

6 Replies to “Data to Decision making”

  1. GOOD INFORMATION WITH REGARD TO DATA ANALYSIS. IT IS ALWAYS CONFUSING WHICH STATISTICAL TOOL TO APPLY FOR WHICH TYPE OF DATA. GOOD TO SEE THIS BLOG. IT ACTS AS A GUIDE FOR RESEARCHERS LIKE US. THANK U SHANKAR SIR.

Leave a Reply

Your email address will not be published. Required fields are marked *