Volume mention forecasting enhances the Mentions Trend widget in Explore by adding predictive insights. Instead of only viewing what has already happened, you can now see projected trends based on historical data. This forward-looking view helps you anticipate changes, spot signals earlier, and move from analysis to action with greater confidence.

Note: This update is currently being rolled out in phases. If you do not currently see it in your account, expect to have access in the near future. 

This article will cover:

Understanding Predictive Analytics

Forecasts in Explore are generated using historical data and advanced time-series models:

• AutoARIMA: Spots time-based patterns, such as weekly cycles or seasonal dips.
• NPTS (Non-Parametric Time Series): A flexible model that works well with irregular or noisy data, since it doesn’t rely on strict mathematical assumptions.

Predictions become available once there’s enough past data (usually a few weeks) and are refreshed automatically as new data is added.

Meltwater uses different models because different types of data can show very different patterns. For example:

• A brand’s social engagement may spike every Friday as people talk about weekend plans.
• A product’s sales may rise gradually over time without any clear seasonal trend.

Using Predictive Analytics in Explore

The Mentions Trend widget in Explore now includes a Prediction button.

1. Click Explore in the left-hand navigation bar
2. Open an existing search or create a new one
3. Navigate to the Mentions Trend widget
4. Click Prediction

5. The volume predictions will show two days of forecast.