Perform forecasting with FBProphet, LSTM with Keras, and Holt's Method on single time series.
Forecasting plays an important role across various fields and industries, including but not limited to Manufacturing, DevOps, and IoT. Forecasting enables better production planning, capacity planning, as well as predictive maintenance.
FBProphet or Prophet is a forecasting time series library. The Prophet frames the forecasting problem as a curve-fitting exercise. It is an additive model where non-linear trends are fit with yearly, weekly, and daily seasonality, plus holiday effects. Prophet is best suited for time series data with strong seasonality. Ideally you want to fit an entire cycle of historical data or several seasons of historical data.
LSTM or Long short-term memory is an artificial recurrent neural network (RNN). Understanding LSTM Networks is a fantastic resource to those wanting to learn more about how LSTMs work.
Statsmodels is a Python module that provides functions for many statistical models (including forecasting and anomaly detection) and statistical data exploratin. The statsmodels Holt's method is used to generate forecasts in the Holt_Method.ipynb notebook. Holt's method is also known as double exponential smoothing. Double exponential smoothing is an exponentially weighted average of the trend and values of the data together. I encourage you to look at Machine Learning with the Telegraf Execd processor plugin to learn about how to implement this forecasting method with the Telegraf Execd processor plugin.