Training and testing the model

To predict new values using the data that we have, we will follow these steps:

• Use 80% percent of the data to train and generate the linear regression model.
• Give the upper and lower models ranges of uncertainty. Remember that a model is just a trend and approximation of the prediction values.
• Test the linear model with the remaining 20% of the data and see how the model fits with these expected values.
• Use the model to predict new values using unknown data.

Build the linear model formula with 80% of the data in Figure 10.19:

Figure 10.19 – Linear regression coefficients for building a formula model

Writing a linear regression model formula

Using the coefficients of the linear model, we build the formula for the regression line:

Intercept B0 = -34.8789
B1 material rotation = 0.63
B2 online marketing = 15.699
Predicted sales revenue from model = Intercept B0 + (B1 * material rotation)...