This article is by Timothy Lam and originally appeared on the Alteryx Data Science Blog here: https://community.alteryx.com/t5/Data-Science-Blog/Expand-Your-Predictive-Palette-III-I-Sales-Forecast-with-Prophet/ba-p/504651
At a high-level, forecasting techniques can be broken down into three main categories:
Historical Average with Sliding Windows
Examples: Seasonal Decomposition, Exponential Smoothing (ETS)
Pros: Simplicity: any tool can integrate like Excel & Tableau;
Cons: Laggardly reaction to changes & overly responsive with outliers, only works with simple structure
Examples: ARIMA, VAR(Good for multiple time series)
Pros: Works with consistent variation, such as established seasonality trends
Cons: Proper assumption on stationarity and homoscedasticity (statistics term for consistent variance/error). Careful predictor selection to prevent multicollinearity (statistics term when predictors are linearly correlated).
Examples: Prophet, GARCH (generalized autoregressive conditional heteroskedasticity), Deep Learning
Pros: Discovering non-linear relationship & data drift in your data.