Forecasting in Zoho Analytics allows you to project future data trends based on historical patterns. By applying statistical and machine learning models directly to your reports, you can generate forward-looking projections that inform planning decisions, set realistic targets, and identify potential issues before they arise.

Prerequisites

Before enabling forecasting on a report, ensure the following conditions are met:

• Your dataset contains at least 7 historical data points for the forecast to be statistically meaningful.
• The report includes a Date or Date-Time column on the X-axis to provide the time dimension required for trend analysis.

Supported Report Types

Forecasting can be applied to the following report types in Zoho Analytics:

• Line Charts
• Bar Charts
• Scatter Charts
• Area Charts
• Combo Charts
• Pivot Tables

Forecasting Models

Zoho Analytics offers several statistical models to generate forecasts. Each model is suited to different types of data and patterns:

• ARIMA: AutoRegressive Integrated Moving Average, suited to stationary time series data with autocorrelation patterns.
• STL: Seasonal and Trend decomposition using Loess, ideal for data with strong seasonal cycles.
• ETS: Error, Trend, Seasonality model, effective for data with exponential trends and seasonal variation.
• Regression: Linear regression-based projection, useful when trends are broadly linear.
• VAR/VARX: Vector AutoRegression models for multivariate time series forecasting involving multiple interdependent variables.

Confidence Levels

When displaying a forecast, you can choose a confidence interval ranging from 70% to 95%. This band represents the statistical range within which actual future values are expected to fall, giving you a visual sense of forecast uncertainty.

Automatic Updates and Data Snapshots

• Automatic Refresh: Forecasts update automatically each time the underlying dataset is refreshed, ensuring your projections always reflect the latest data.
• Data Snapshots: You can take snapshots of forecast data at a point in time to preserve records for comparison and reporting purposes.