Advanced prediction strategies
1. Advanced prediction strategies
Weighted averages are fairly straightforward to calculate, especially when we have intermediate measures. Now we'll turn our attention to a more advanced way of predicting unobserved values using linear models.2. Linear models
Linear models are algebraic equations that allow us to predict unobserved values using other measures or data from other individuals. You'll often hear people talk about these models as "lines of best fit." For example, imagine we have data on individuals' height and weight and want to predict the weight of another individual based on this set of data. In this new carnival weight-guessing game, we would use the relationships between the height and weight of other people to guess a new person's weight. Linear models define the relationship between the dependent variable or variables, which are the factors that cause changes in the independent variable or outcome.3. Forecasting
The FORECAST() function in Spreadsheets allows us to make these predictions using linear models. It takes three arguments: the x value of the dependent variable which we want to use to predict (in this case, we'll use a period 2 grade of 6 to predict the student's final grade), the y data, which are the actual outcomes (in our case, final grades in column A), and the x data, which represent the set of predictors or levels of the dependent variable we mentioned earlier (these are the period 2 grades in column C in this example).4. Forecasting
In this case, we want to predict the final grade based on the x value of 6, which is the period 2 grade for the first student.5. Forecasting
Our data_y values, or outcome data are in column A.6. Forecasting
And our predictor, or data_x, values, are in column C.7. Forecasting
So we predicted a final grade of 3.88 for a student who earned a Period 2 grade of 6. As we can see, this model underestimates this student's final grade in this case. We will discuss prediction accuracy later, but we should note that we don't expect our models to be 100% accurate, since they can't account for all the messy variation that exists in the real world.8. Let's practice!
Now it's your turn to peer into the future using the FORECAST() function. Let's get some practice in the exercises that follow.Create Your Free Account
or
By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.