This model captures fine details, which is great, but it is neither convenient nor easily scalable.

First, let’s place all the inputs/feature variables on the far left.

Use arrows to connect all the weighted sum variables to the score, representing the weighted summation of all feature variables.

With the arrows in place, the signs of all the coefficients become less important—go ahead and remove them.

Finally, we use an arrow with an activation function symbol to connect the score to the prediction/output.

This way, we have a diagram that is both visually appealing and convenient for representing our model.