DATA SCIENCE

Data Science Explained

What is data Science?

      The most appealing and renowned word today over the internet is data science, once called "Data the new oil" by Economics. Over the years, data science has gained more popularity than any other field.  Data Science is not one specific field, rather it is a broad term under which multiple fields reside. Data science ultimately consists of fields such as software engineering or computer programming, business studies, statistics, mathematics and most importantly the domain knowledge in which the data science is being applied to, collectively forms data science.

      Data science includes multiple processes like storing data; creating pipelines or extracting data also known as ETL(extract transform load), cleaning or preparing data for analyzing and visualization, and finally for the model building. Above all, the first thing it starts from is by setting up the business goal. Once the business goal is predetermined than it provides meaning to the dataset. Without the business goal, the data set has no meaning to it.

         Data science is used to carry out studies of the large volume of data present in the form of tables and observations. Data science is the study of evidence recorded in the form of observations. It is help full to get insight out of the data and the key findings reduces the amount in decision making. 

What is the purpose of Data Science?

    Data science is used for analytics and predictive modeling etc. Today, data science is used for predicative, diagnostic and perspective analytics. Data science is essential for modern businesses and companies. Using machine learning models, companies can forecast future sales, and all these are just the initial step of data science. The main goal is to create an ultra refined data product. The data product must be new for some people or for those who are beginning in data science. Data science plays and vital role in decision making, data science roles vary from industry to industry if you are applying data science in business then one of the most widely used task is to forecast the future sales, and if you are using data science in the health care sector than it determines whether the patient has a disease  or not.

Why Data Science is important?

        In todays world, data science gives new competitive edge to the businesses around the globe. Large volume of data stored on cloud, hard drive, and else where is not irrelevant instead it is new driving force for business, engineering and research development. Tomorrows success is built on past data. Business intelligence, Business analyst, Engineering, Health Care and Security, you name it, data science serves as the main part of any industrial sector. It helps you decide the risk and guarantees you success in the long run. Today, all decisions are data driven or what is commonly known as "Data Driven Decisions."

What is Data Product?

           People are mostly familiar with the term product where there is a manufacturing plant, but we, as data scientists, create and build models and evaluate them up to their optimal level, which is known as data product. The process of making a data product is simple. A company has a data set in a certain format i.e. csv, excel, and json a data scientist simply applies a machine measure algorithm to create a model which is then deployed on a cloud or local host. This model is known as a data product

What are the Application of Data Science?

         There are countless data science applications present today. Data science applications can be seen in security, health, business, defense, government, financial, weather, aerospace, and geo-spatial etc. The task may vary from industry to industry. If data science is implemented in the financial sector, then the task would be regression, and if it is in the health care sector, the task may be classification. The following is the list of data science applications with respect to industries.

Defense industries: Robotics and modern day weapons are all equipped with machine learning algorithms. Deep learning and artificial neural networks are the brains of robots used in the defense sector. Computer vision, for facial recognition around the globe, people use convolutional neural network or CNN.

Health Industry: Classification is the most used technique of data science in health care. Classification is a widely used technique to classify whether the patient has the disease or not. Classification is also used in image recognition. Multiple CT scans and X-rays determine whether there is an injury or not using image processing or computer vision.

Automotive: Based on the data whether the car is serviced or not, is it going to break down or not? This is achieved using a technique called logistic regression.

Business: We can forecast the future sales of a company. Using regression, we can forecast future sales, this is one of many data science techniques implemented in business. In business, we can also forecast future sales using time series. The Time series is one of the widely used techniques in business.

Government: Data science also helps governments to predict what could be the future GDP of the country. This is also achieved with the help of regression. Time series, of growth and using geospatial data to keep record where the most crimes happen.

Sports: The prediction of a score in any game is achieved using regression. With the help of regression, you can predict which team is going to win. There is more than one data science technique used in the field of sports. Many teams, based on previous data, implement classifications to determine whether the side is going to win or not.

Data Science and Data Analytics.

    Data analytics is a subdomain of data science in which you analyze the data set or any given raw data. Data analytics is among the first step for getting insight from the data. Data analytics is used to find out the trends and insight from the raw data. Data analytics is done to make data talk to you. The data talks about itself once you start visualizing it. Sometimes, data science and data analytics are used interchangeably. It is because both are done to make data speak its story. The key difference between data science and data analysis is that data science is used in broader terms and includes all aspects of data, while data analytics scope remains limited and has nothing to do with the operational activities regarding data, i.e storing, cleaning and ETL etc. In data analytics you ask several questions and make try to find out the answer by visualizing it.

Data analytics is a process in which statistics of data and visualizing of data takes place. Data analytics mainly consists of the following three processes. Each three of them answered three main questions.

Descriptive analytics.

In descriptive, analytics we answer the question what happened?

Diagnostics analytics 

In diagnostic, analytics we find out the answer of why it happened?

Predictive analytics

In predictive, we try to determine what could be the possible outcome simply by answering the question what will happen?

Perspective analytics

In perspective, data scientist search for the solution and made suggestion on behalf of the predictions made and explains them. This is the type of analytics in which a data scientist determines the solutions as per the out come from the previous steps of analytics. 

Data science and machine learning

       Data science and machine learning are mostly considered the same things. But this is not the case, since machine learning is the last step which comes after several preprocessing steps like ETL(extract transform load), data cleaning, EDA(exploratory data analysis), and data analytics. The last step in data science is machine learning..

     In machine learning,  the models learn and train themselves from the given data. Machine learning is a  domain learning from the data but explicitly  limits to data modeling, with the help of prescribed algorithms. Machine learning algorithms are used on the basis of, which type of learning is offered. Mainly, there are three types of machine learning; supervised, unsupervised, and reinforcement learning. The type of machine learning is based on the type of data, a data scientist processes, and what  the target columns are. There is also a hybrid type of machine learning called semi-supervised machine learning.

 

Author: MUHAMMAD SAAD