What it needs for successful implementation of a Data Science project
From past few years there is a lot of buzz in the market about data analytics and data science. However, when it comes to on time successful implementation of an end to end data science project, several times it either fails or does not end up with the expected results. Many asks me what is needs for driving a successful data science project.
I want to keep this article crisp and short and hence, based on personal experience, the key things needed for driving a data science program can be summarized in the following diagram:
Passion + Knowledge about business interpretation of data : Analyzing and interpreting data may be frustrating at many times, especially datasets with huge attributes and high cardinality. Hence passion to deal with data plays an essential role. Passion along with application of statistical techniques to analyze data is pivotal. However, it will bring the value only when the interpretation and inferences are made keeping business in mind, otherwise it will always remain limited with a technical flavor with little value for the business users - thereby reducing its chances of success.
Knowledge to identify proper models and algorithms based on the Use Cases : Right skills and knowledge to understand the requirements, and identify the right set of ML Models & algorithms for those key use cases is a pillar to success. ML is a huge area and there are innumerable techniques and algorithms present, however the efficiency to identify the right ones to correctly fit the requirements will decide the quantum of success. A failure in this part would lead to trial & error by experimenting with different models, resulting in loss of vital time of the project.
Right tools and software: This is third in the priority because, this is going to bring in the necessary value only when the first two steps are executed properly. One important thing to keep in mind is that Data Science is not Python, Tableau, Hadoop, Azure etc. - all these are technology levers to successfully implement the above two steps. Hence unless the first two steps are conceptualized and planned properly, the tools and software will remain like unutilized weapons.
However, the most critical piece in all these is something else - which is putting loads of Common Sense in all the phases. There are several operational issues and challenges which can be resolved in each of the above phases by putting enough common sense. Data Science techniques, ML algorithms or tools will not be able to produce optimum results, unless enough common sense is put through to make them work in the most efficient way. Hence, assume this to be the backbone supporting the entire end to end implementation throughout.
Visit my Facebook Page @ facebook.com/FBTrainBrain/
Read my other article
