Why study data science?
An estimated shortage of up to 190,000 data analysts in the U.S. is creating high demand for data scientists with the know-how to use data to make effective decisions. From predicting consumer behavior to extracting information from medical images, you will graduate ready for a dynamic career that inspires global change.
Why study data science at ASU?
Modern science and technology use sophisticated mathematical and computational tools to extract patterns from large, complex and often unordered data sets. Machine learning and data mining are invaluable technologies with applications as diverse as detecting fraudulent online credit-card transactions, understanding the dynamics of social movements, and personalizing medical treatments based on a tumor's unique genetic profile.
ASU's BS degree program in data science prepares students to be critical analysts and users of data in a variety of areas such as business, research and government. This transdisciplinary program allows students to choose a focus area from a variety of fields to center their understanding of data science. With a mathematical core consisting of linear algebra, statistical inference and classification, data mining, machine learning and associated computer methods, students leave the program with a strong background in data-related skills that are useful in solving real world issues.
A typical workflow in many data science projects can be summarized in the following diagram.
Capturing Data and Data Cleaning, Munging, Wrangling
Collecting large amounts of data is a challenge by itself. It may involve querying data from web servers, databases, API’s, online repositories, etc. Data are often noisy and almost never clean. There are often missing data, or data incorrectly captured or stored. Also, not all data are equally useful and some cannot be used directly. They may need to be modified and transformed into more useful datasets. These are all challenges that have to be dealt with before cleaned and modified data can be used as an input in mathematical and statistical models. Dealing with these challenges became a part of data science, at least in its broader sense.
Exploratory Data Analysis (EDA)
After cleaning and modifying data, we would like to use them in statistical models in order to make predictions about the business or domain from which the data came. But, in order to determine which statistical models are appropriate to apply or which hypotheses about the business are reasonable to assume and test, it is important to get a first intuition about the data and see what they are telling us. For that purpose, we perform exploratory data analysis. We often visualize data by plotting various graphs, histograms, bar charts, pie charts etc., with different features/variables/columns. This helps us get first impression and intuition, and discover signatures of interesting phenomena the data are telling us about.
Mathematical and Statistical Modeling; Final Analysis
After visualizing data and getting intuition about what they show us, we can often raise some questions that need answers, and hypothesize certain claims about the data that need to be tested. We would also like to make predictions on future data. This is all done using mathematical and statistical modeling. That is, applying math and stats tools and methods on the cleaned data as inputs. This often requires an intensive programming/coding.