Understanding Our Forecasting Tool

Forecasting: Understanding Our Forecasting Model

Our forecasting model is designed to provide you with personalised predictions that enable informed decision-making. By forecasting future metric values, you can anticipate trends and prepare your strategy with a clear view of what lies ahead.

How It Works

Learning from Past Data:

 The model begins by analysing your historical data, establishing what “normal” looks like for your metrics. This long-term view forms the baseline for forecasting.

Emphasising Recent Trends:

Similar to our anomaly detection model, the forecasting model places extra focus on recent data. By concentrating on the latest weeks, it captures seasonal fluctuations and emerging trends to ensure the forecasts are up-to-date.

Customised Model Fitting:

 In the background, our forecasting algorithm tests various models against your data. It continuously selects the best-suited model for your specific metric, ensuring a customised approach. Running every day, the algorithm keeps learning and evolving, meaning you might see a more accurate forecast tomorrow compared to today.

Providing Confidence Intervals:

 Each forecasted value comes with an upper threshold and a lower threshold. These boundaries offer a clear range within which the future metric values are expected to lie, giving you valuable context for planning.

Why This Matters

With personalised forecasts at your fingertips, you gain a proactive edge in planning and resource allocation. By anticipating future trends, you can mitigate risks and seize opportunities before they fully materialise. This feature empowers you to adjust strategies in real time, ensuring that your decisions are always data-driven and forward-looking.

Important to Note

Because forecasts are based on predictions, they may not always be 100% accurate. There’s always a slight margin of error, so it’s essential to take this into account when reviewing forecasts. While our model is designed to be as precise as possible, occasional deviations can occur. Additionally, forecasts are currently available for a limited set of metrics and extend up to one month into the future.