Data Science Methodology-101
Steps to follow while being a Data Scientist to solve any specific problem
In this reading, we will discuss the steps that are the real need for a Data Scientist while he/she going to solve any real-world problem.
In this, we discuss the five stages through which a Data Scientist must be passed out to obtain an optimal solution for the required problem.
Here is the hierarchy of the steps to be followed:
The following are the 10 steps cycle in which we take two-barrel at a time to discuss and evaluate parallelly.
1- Problem To Approach:
This is the first step that includes two components namely Bussiness Understanding & Analytic Approach. So, let us first discuss the Bussiness Understanding stage. In this the DS(Data Scientist) first considers the consequences of the Bussiness for which he going to provide the solution for a problem, it means in this he evaluates the problem with reference to the business point of view. Now after this step the main thing to apply the approach to solve any specific dilemma, for this the second step takes place here which considers the Analytical Approach to let him know which approach is to be used according to the need. The Analytical Approach includes three approaches namely, Descriptive Approach(for current status/statistics), Predictive Approach(when you have to work for future insights), Classification Approach(when you supposed to give an answer in True/False or Yes/No)
2- Requirement To Collection:
This stage refers to the stage where DS must look for the requirements of data that he needed to step ahead in his work but for this, the main thing is, even he has configured the requirements for data that uses to solve the problem but if he really gets those form of data that actually going to help him. These are two different things that what is your requirement and what you actually collected. So, what you need is to determine the real need and collect the relevant data for your sake.
3- Understanding to Preparation:
So, now when you understand the business requirement and also analyze that what approach is to used and also collected the related data for your problem now, the stage is to explore the real insights, discoveries and reveal those treasure from the data that actually matters rather than anything else is called Understanding stage and for the Preparation, the one must have to assemble the data(derived after Understanding stage) for the further implementation in the model.
4- Modelling to Evaluation:
After going through the stages that described above now, it’s time to make the actual Model that suppose to solve the problem for which you follow all the necessary steps but one main thing is to mention here is that the Model must fit the actual need of the business requirement from where it all starts, now you can evaluate your model into action but for this, it is necessary to test a model first with some small bunch of related task(which is called testing) for which it is designed for, rather run through it in a fully complex environment without being tested
5- Deployment to Feedback:
This is the step that is supposed to be very interesting regarding all the processes that we discussed earlier in this article is, to Deploy your model to the business problem and prepare yourself to be ready for the Feedback stage for getting either positive or negative result for your efforts. As this is a continuous stage which means you surely have made to amendments to the model according to the feedback and a good DS always looking for feedback so he can make his model much more efficient as before
Conclusion:
Woohoo🎉!!! now as you covered all the necessary steps that a DS must need to start over a specific scenario(problem), one main thing to discuss is that all the 10 stages in this model are iterative in nature, means you can go and forth all the stages repeatedly as long as you can not find the optimal solution for the problem
