Following the previous article on Understanding of Historical Sales With Time Series Analysis, which was primarily focused on the analysis of time series, today, we shift our focus to the exciting topic of Time Series Forecasting.

To recap, KPI forecasting is a hot topic in digital analysis because it provides valuable insights for decision-making, particularly in anticipating future trends like upcoming sales. This aids in managing inventory and adjusting marketing budgets across channels, ultimately driving sales growth.

Time Series Forecasting involves the use of powerful and complex algorithms like ARIMA (AutoRegressive Integrated Moving Average) to capture patterns in data. Additionally, regression models establish relationships between the dependent variable (e.g. sales) and independent variables (e.g. marketing spend, weather), enabling businesses to consider external influences for a more comprehensive understanding of sales fluctuations. By combining ARIMA and regression, known as ARIMAX, the approach leverages both time series patterns and external factors for accurate predictions. This makes it particularly suitable for scenarios like forecasting sales while considering marketing activities across various channels.

When future marketing spends and other external factors are provided to the model, it generates sales forecasts based on the planned influences (see figure 1).

Moreover, by decomposing historical sales into influencing factors (see figures 2 and 3), businesses gain valuable insights for fine-tuning these factors to achieve desired sales. This approach proves extremely helpful in crafting effective strategies for reaching sales goals based on past learnings.