1. How clear is your crystal ball?
We have now covered a couple ways we can make predictions, but how do we know they are accurate? This lesson will highlight ways we can evaluate the accuracy of the predictions we make.
2. Predicted vs. actual
We measure prediction accuracy by comparing the values we forecast to the actual observed outcomes.
For example, we can subtract students actual final grades from the values we predict in our forecasts.
This lets us know how far off our predictions are.
3. Absolute deviation
But taking all these differences could mislead us if we don't make an important correction. After all, some of our predicted values will be higher than the actual observed value, while others will be lower.
By using the ABS() function to take the absolute value, we calculate the total difference between the predicted value and the true outcome, which allows us to assess the accuracy of our predictions.
We call this difference the absolute deviation.
4. Mean absolute deviation
Once we have calculated the absolute deviation for each observation, we can take the
average of these values to get a sense of the overall accuracy of our predictions.
This can be especially useful for comparing different sets of predictions,
as the one with the lower absolute deviation will typically yield more accurate predictions.
5. Let's practice!
Now that you have some tools to compute prediction accuracy, let's see how accurate our predictions are!