Jan 18, 2019

Understanding a Time Series Analysis

There are many specialized tasks and skills in the field of data science, business analysis, and machine learning, and forecasting using time series analysis is one you might encounter if you choose a career in one of these areas. A time series is a set of data points collected during a period of time at regular intervals rather than not randomly. Some everyday examples of data that people can collect in a time series include ocean tide heights, weather events, and even demand for and pricing of products. Data scientists can use time series analyses to interpret a wide range of types of data for many different purposes like helping companies make predictions about retail trends or business forecasts. Learn what's involved in time series analysis to see if this side of data science piques your interest.

Time Series Analysis Basics

There are two general goals of performing a time series analysis: looking at data to interpret what happened in the past and using the data to predict what's likely to happen in the future. It's sort of a crystal ball, but one that's based on months' or even years' worth of concrete information. Data scientists search for patterns in the data that can apply to future time periods. They describe the relationships between the data and both qualify and quantify those patterns. Once this process is complete, a data scientist can compare the patterns and data to other semi-related sets of data, which allows them to make predictions that have a higher level of accuracy.

Almost every industry that exists today can use time series analyses in some way. Within the fields of data science and machine learning, these analyses are commonly used in exploratory data mining, statistical analysis, pattern recognition, and classifications with training sets. Professionals who deal with data can also use a number of different tools to investigate and interpret data. An understanding of different visualization methods, including overlapping and separated charts, is also essential for performing a time series analysis. Time series analyses also frequently make use of models when interpreting data.

What's the Role of Models?

In the context of time series analysis, a model is a system that helps people make forecasts. Data scientists and other professionals typically use error, trend, and seasonality (ETS) or autoregressive, integrated, moving average (ARIMA) models to compile data and use it to make predictions. ETS is effective for data that shows trends or seasonal cycles, and ARIMA is preferable when data comes from the same source and clearly shows a past-present correlation. It's often helpful to run both models on data, regardless of its patterns.

If you're considering becoming a data scientist, it's important to understand how to perform a time series analysis. To learn about other skills related to the data science field, browse the Udacity catalog and discover all the pathways data scientists can take in their careers.