The weights w11, w12,... w23 and biases b1 & b2 can also be represented as matrices, initialized as random values. The first row of w and the first element of b are used to predict the first target variable, i.e., the yield of apples, and similarly, the second for the yield of oranges. Our model is simply a function that performs a matrix multiplication of the inputs and the weights w (transposed) and adds the bias b for each observation. Let's create a function called model that will predict output by calculating the above operation when the inputs is passed as a parameter: Now, let's compare the predictions with the actual targets: As you can see, the difference between our model's predictions and the actual target values is vast. So, we need to improve our model to reduce the difference.