Schema Performance Analytics' forecasting feature uses machine learning to provide performance forecasts for your website's clicks, impressions, and click-through rate based on historical trends. This article describes the Forecasting feature and it's applications and constraints.
TABLE OF CONTENTS
- Forecasting Features and Applications
- How To: Apply Forecasting to a Graph
- How To: Edit Forecast Properties
Forecasting Features and Applications
- Multiple levels of seasonality (weekly and quarterly trends)
- Anomalies are automatically excluded
- Missing values are imputed
Forecasting can be applied to any line graph in Page Level SPA reports.
Presently, Forecasting is applied to the "Clicks and Impressions by Date" graph in the following Page Level SPA reports:
- URLs with Schema App Markup
- URLs without Schema App Markup
- All URLs Performance
How To: Apply Forecasting to a Graph
Step 1: Go to any line graph in Page Level SPA Reports.
Step 2: Select the Menu options hamburger icon.
Step 3: Select Add forecast.
Step 4: Review Forecast and adjust Properties as needed
Schema Performance Analytics automatically analyzes the historical data using machine learning, and displays a graphical forecast for the next 14 periods. Forecasted data will appear in yellow.
Users can make adjustments to the following properties to tailor the forecast to their needs
- Forecast Length
- Prediction Interval
- Forecast Boundaries
How To: Edit Forecast Properties
The Forecast properties panel can be used to customize one or more of the following settings. To modify
A length of a period will depend on the period of the graph. In the example used below, the periods are equivalent to days.
Input the number of periods into the future you would like to predict.
Input the number of periods into the past you would like to review. Use this to look for patterns to base the forecast on.
Prediction interval is an estimate of an interval in which future observations will fall with a certain probability given what has already been observed.
Changing this value will set the estimated range for the forecast and change the width of the band of possibility around the predicted trend lineline. Increasing the Prediction Interval will increase the range of possibility, and lowering the Prediction Interval will decrease the range of possibility.
Example: Data Forecast with a Prediction Interval of 30.
Looking at data for August 8th shows a band of possibility (upper bound and lower bound) that is quite narrow.
Example: Data Forecast with a Prediction Interval of 90
Looking at data for August 8th shows a broader range of possibility (upper bound and lower bound) for each forecasted data point.
Seasonality is a characteristic of a time series in which the data experiences regular and predictable changes that recur every calendar year. The default setting, Automatic, detects seasonality for you.
This can be modified by applying values from the given range of 1-180. For example, if you notice a weekly pattern in your data, and your data is provided in daily granularity, set your seasonality to 7. If your data has a quarterly pattern, and your is provided in monthly granularity, set your seasonality to 3.
Set a minimum and/or maximum forecast value to prevent forecast values from going above or below a specified value.
Was this article helpful?
Thank you for your feedback
Sorry! We couldn't be helpful
Thank you for your feedback
We appreciate your effort and will try to fix the article