AWS Security ChangesHomeSearch

AWS forecast documentation change

Service: forecast · 2026-05-04 · Documentation low

File: forecast/latest/dg/metrics.md

Summary

Updated image descriptions and alt text for mathematical formulas and visualizations related to forecast accuracy metrics.

Security assessment

Changes only involve improving image accessibility and descriptive accuracy of technical formulas/diagrams. No security vulnerabilities, configurations, or features are mentioned or modified.

Diff

diff --git a/forecast/latest/dg/metrics.md b/forecast/latest/dg/metrics.md
index 80ca0c123..ed78017ce 100644
--- a//forecast/latest/dg/metrics.md
+++ b//forecast/latest/dg/metrics.md
@@ -99 +99 @@ The loss function is calculated as follows.
-![Mathematical equation for weighted quantile loss function with tau parameter.](/images/forecast/latest/dg/images/metrics-quantile-loss.png)
+![Formula for weighted quantile loss with tau parameter, max functions, and summations.](/images/forecast/latest/dg/images/metrics-quantile-loss.png)
@@ -125 +125 @@ In retail, the cost of being understocked is often higher than the cost of being
-![Two probability distribution curves showing demand forecasting at P50 and P75 quantiles.](/images/forecast/latest/dg/images/p50-p75-prediction.jpg)
+![Probability distributions showing 50% probability at P50 and 75% probability at P75 demand levels.](/images/forecast/latest/dg/images/p50-p75-prediction.jpg)
@@ -139 +139 @@ When the sum of observed values for all time points and all items is approximate
-![Mathematical formula for WAPE showing summation of absolute differences divided by sum of absolute values.](/images/forecast/latest/dg/images/WAPE.png)
+![WAPE formula showing sum of absolute differences between actual and predicted values divided by sum of actual values.](/images/forecast/latest/dg/images/WAPE.png)
@@ -154 +154 @@ Amazon Forecast previously referred to the WAPE metric as the Mean Absolute Perc
-![Mathematical equation showing the equivalence of wQL\[0.5\] and WAPE\[median\] metrics.](/images/forecast/latest/dg/images/wql-to-wape.PNG)
+![Mathematical formula showing wQL\[0.5\] equals twice the sum of absolute differences divided by sum of absolute actual values.](/images/forecast/latest/dg/images/wql-to-wape.PNG)
@@ -160 +160 @@ Root Mean Square Error (RMSE) is the square root of the average of squared error
-![Mathematical formula for Root Mean Square Error \(RMSE\) with summation and square root.](/images/forecast/latest/dg/images/metrics-rmse.png)
+![RMSE formula showing square root of sum of squared differences between predicted and actual values.](/images/forecast/latest/dg/images/metrics-rmse.png)
@@ -181 +181 @@ Mean Absolute Percentage Error (MAPE) takes the absolute value of the percentage
-![Mathematical formula for Mean Absolute Percentage Error \(MAPE\) calculation.](/images/forecast/latest/dg/images/mape.png)
+![MAPE formula showing one over n times the sum of absolute value of A sub t minus F sub t divided by A sub t.](/images/forecast/latest/dg/images/mape.png)
@@ -200 +200 @@ Mean Absolute Scaled Error (MASE) is calculated by dividing the average error by
-![Mathematical formula for Mean Absolute Scaled Error \(MASE\) with summation and absolute value notations.](/images/forecast/latest/dg/images/mase.png)
+![MASE formula showing mean of absolute errors divided by scaled mean absolute error.](/images/forecast/latest/dg/images/mase.png)
@@ -270 +270 @@ During the **Create predictor backtest export** stage, set the **Export name** ,
-![Form for exporting predictor backtest data to S3, with fields for name, IAM role, and location.](/images/forecast/latest/dg/images/console-export-screen.PNG)
+![Create predictor backtest export page with fields for export name, IAM role, and S3 location.](/images/forecast/latest/dg/images/console-export-screen.PNG)
@@ -287 +287 @@ Quantiles can provide an upper and lower bound for forecasts. For example, using
-![Graph showing forecast quantiles with P99, P90, P50, P10, and P1 lines over time.](/images/forecast/latest/dg/images/quantiles-intervals.png)
+![Time series forecast showing P50 median with shaded confidence intervals for P10, P90, P1, and P99.](/images/forecast/latest/dg/images/quantiles-intervals.png)
@@ -312 +312 @@ For example, to create a predictor using the `0.01`, `mean`, `0.65`, and `0.99`
-![Form for entering forecast types with fields for type names and quantile values between .01 and .99.](/images/forecast/latest/dg/images/predictor-custom-quantiles.png)
+![Four forecast type fields with values .01, mean, .65, and .99 entered respectively.](/images/forecast/latest/dg/images/predictor-custom-quantiles.png)
@@ -324 +324 @@ The backtest window must be at least as large as the forecast horizon, and small
-![Graph showing training and testing periods for four backtest scenarios over time.](/images/forecast/latest/dg/images/evaluation-backtests.png)
+![Four backtest scenarios showing training and testing periods between start and end dates.](/images/forecast/latest/dg/images/evaluation-backtests.png)
@@ -351 +351 @@ For example, to run 2 backtests with a testing set of 10 time points, set the fo
-![Input fields for number of backtest windows and backtest window offset with example values.](/images/forecast/latest/dg/images/predictor-backtest-windows.png)
+![Number of backtest windows field set to 2 and Backtest window offset field set to 10.](/images/forecast/latest/dg/images/predictor-backtest-windows.png)