Time Series Forecasting: ARIMA vs LSTM vs PROPHET
Time Series Forecasting with Machine Learning and Python
Summary
The purpose of this article is to find the best algorithm for forecasting, the competitors are ARIMA processes, LSTM neural network, Facebook Prophet model. I will walk through every line of code with comments, so that you can easily replicate this example (link to the full code below).
We will use a dataset from the Kaggle competition “Predict Future Sales” (linked below) in which you are provided with daily historical sales data and the task is to forecast the total amount of products sold. The dataset presents an interesting time series as it is very similar to use cases that can be found in real world, as we know daily sales of any product are never stationary and are always heavily affected by seasonality.
Full Code (Github):
Dataset (Kaggle):
