Time series Forecasting - ARIMA and R

Time Series Forecasting is the ability to predict future events based on learning from historical data (time series). The component of the time brings in concepts of trends, cycles, seasonality to be accounted for in future predictions. The forecasting techniques are applied to sales forecasting, demand forecasting, liquidity forecasting, budget forecasting, inventory forecasting, predictive maintenance, and many other examples that many of us can relate to in every business domain.

This is a three-part series consisting of the following sessions

1. Time Series Forecasting using ARIMA
2. Time Series Forecasting using Neural Networks
3. Explain the machine made forecast prediction

Objective: In this meetup we will discuss the fundamentals of forecasting relying on time series methods, mostly focusing on ARIMA (capabilities, advantages, limitations). Hands-on exercises will help participants understand how to use tools, available in R language, to build models on their own.

Agenda

1. Problem description - Rossman Sales Forecasting (Kaggle example)
2. Fundamentals of ARIMA (Autoregressive, Integrated, Moving Average)
3. Hands-on exercises
4. Understanding results using the shiny app
5. What are the limitations of ARIMA solution?

Speaker: Sergey Samsonau, Data Scientist, airisDATA