Statistics for Big Data For Dummies

Chapter 17

Using Your Crystal Ball: Forecasting with Big Data

In This Chapter

Understanding the properties of time series

Transforming data to fit modeling assumptions

Forecasting a time series using ARIMA modeling

Seeing how simulation is used for forecasting purposes

You can use a few different techniques to forecast the future values of a time series:

• Time series regression
• ARIMA modeling
• Simulation

Chapter 16 covers time series regression. This chapter covers ARIMA modeling and simulation techniques. ARIMA models use the past values of a time series to develop a forecasting model, whereas simulation techniques are based on a statistical model of the variable that’s being forecast.

ARIMA Modeling

ARIMA (autoregressive integrated moving average) modeling uses the past behavior of a time series to determine its key statistical properties and takes this information to develop a forecasting model.

ARIMA modeling is only valid for a time series that’s both stationary and nonseasonal. A time series is stationary if the basic statistical properties of the time series don’t change ...