Accelerating Time Series Analysis - Use Cases, Types, Methods, and Top Python Libraries

Time-series data is a sequence of data points collected over successive time intervals, allowing us to track changes over milliseconds, days, or even years. Time-series data is commonly used in various fields such as economics, finance, healthcare, engineering, and more. The use cases are numerous, but it's generally used to analyze patterns, trends, and behaviors that evolve over time.

Example Use Cases

Financial trading systems: stock prices, classic securities, newer cryptocurrencies.
Weather patterns: temperature, dew point, humidity, pressure, etc readings at the local, regional, or global level to monitor or predict future weather trends and events.
Economic Indicators: monthly unemployment rates, consumer price index (CPI), job listings, etc.
Healthcare patient monitoring: biosensor data for in-hospital patient monitoring and/or remote patient monitoring of heart rate, blood pressure, oxygen saturation, respiratory rate, physical activity, blood sugar levels, etc.
Physical system monitoring: sensor data from equipment, machinery, and connected devices (IoT) for use cases such as status tracking, utilization forecasting, energy consumption, and preventative maintenance.
Software system monitoring: virtual machines, containers, services, applications, and web pages to monitor traffic, utilization, status, logs, etc.
Asset tracking applications: vehicles, trucks, physical containers, and pallets, for dynamic routing, arrival time prediction, and supply chain optimization.
Business intelligence tools: tracking key metrics and the overall health of the business

Types of Time Series Analyses

Univariate Time Series: only one variable varies over time.
Multivariate Time Series: more than one variable varies over time. Each variable not only depends on its past values but also could depend on the other variables.

Methods of Time Series Forecasting

Autoregression (AR): Univariate
Moving Average (MA): Univariate
Autoregressive Moving Average (ARMA): Univariate
Autoregressive Integrated Moving Average (ARIMA): Univariate
Seasonal Autoregressive Integrated Moving-Average (SARIMA): Univariate
Seasonal Autoregressive Integrated Moving-Average with Exogenous Regressors (SARIMAX): Univariate
Simple Exponential Smoothing (SES): Univariate
Holt Winter's Exponential Smoothing (HWES): Univariate
Vector Autoregression (VAR): Multivariate
Vector Autoregression Moving-Average (VARMA): Multivariate
Vector Autoregression Moving-Average with Exogenous Regressors (VARMAX): Multivariate

Below is a list of popular and trending Python packages to help you speed up your time series analysis. There is a lot of overlap between these packages and obviously, we need to understand the capabilities and limitations of both the packages chosen to utilize and the use case for the analysis for a good fit. There are many articles online about the potential controversy with the forecasts from packages such as Facebook's Prophet, but if the only tool you have is a hammer, you tend to see every problem as a nail...