Data Science
“Data Science is what data scientists do.”
Data scientists are the people like us they may be an engineer, statistician, Phd. scholar, business tycoon, finance investor and so on.
Data scientist is the person with the following three features:
1. Curiosity
2. Argumentability
3. Logically skilled.
Data scientist is someone who follows the art of story-telling. He is best at describing the analytical and predictive approach of working with data in the form of a story.
In broader view, data science is the field that is based on analyzing, manipulating and then finding results for the given sets of data in huge amount. Data science is one of the uprising field at present scenario. The future could be witnessed with the applications of data science. We can imagine how it would be like ahead with the help of data science.
Various applications of data science are:
1. Artificial Intelligence
2. Machine Learning
3. Finance
4. Stock
5. Health Care
6. Research
7. Banking sector
and many more.
Data science is more a practice based on the methodology that data scientist practice. Data science methodology includes ten important steps to get results out of the datasets we are supposed to work with.
1. Business Understanding: Business understanding is the foremost step in the methodology of data science. It requires understanding business issues and problems. Identifying business variables needed for prediction.
2. Analytic Approach: Analytic approach is the use of analysis to break a problem into simple steps necessary to solve it. In context of data science analytic approach is analyzing the data to convert it into more understandable format and follow further steps of the methodology.
3. Data Requirements: In data science we need to find the requirements, the kind of data to be processed according to the research we are going to work on.
4. Data Collection: Data collection is one of the important aspect of this methodology. This is the part data is being collected from various sources and resources, communicating with number of people taking in account previous research works. Many people will suggest you never to indulge in this but without collecting data no further action could be taken.
5. Data Understanding : Data understanding comes up with the objective to understand features or attributes of data, gathering some useful information by identifying key characteristics in the data. Most obvious is to find out missing values, outliers and inaccuracies.
6. Data Preparation: Data preparation is referred to as data cleaning phase. It is sometimes quite a boring task as data acquired in the previous steps is not in a usable format. This editing and formatting may involve coding or making a spreadsheet document.
7. Modelling: This is the core step in the methodology. The actual writing, refining and running of programs is performed here. In simple sense, actual model is displayed to analyze and carry out the important aspects of the research.
8. Evaluation: Evaluation step is the mathematical approach. It depends on the model which approach is used. Like if it is spam mails model then average accuracy is considered for evaluation.
9. Deployment: In this phase data science model is deployed in test environment before deploying in the market or in public.
10. Feedback: This is the most important phase in any methodology. Feedback plays a very important role in deciding the success of any research or work done. It will help in improving the model for better results.
With the flowchart it could be seen that the process is never-ending, it is iterating after Feedback, i.e. never-ending process.
Now-a-days data science plays a vital role from little to large utilities we use, i.e. from online shopping recommendation to human activity recognition. This is just one example, there are numerous examples if you look around.