以下代码中使用什么公式来计算ys？

Question

I'm not able to understand the formula for predicting the ys to plot the graph.

How can it be ys = (-theta[0] - theta[1] * xs) / theta[2] ?

fig, axes = plt.subplots(1, 3, sharex=True, sharey=True, figsize=(12, 3))
axes = axes.ravel()

for k, theta in enumerate(tha[:3]):
    ax = axes[k]
    xs = np.arange(0, 1, 0.1)
    ys = (-theta[0] - theta[1] * xs) / theta[2]
    ax.plot(xs, ys, lw=0.5)
    dfa.query('label ==  1').plot.scatter(x='x1', y='x2', ax=ax, color='blue')
    dfa.query('label == -1').plot.scatter(x='x1', y='x2', ax=ax, color='red')

Answer 1

Here you don't predict ys , in the formula both xs and ys are your features, so probably better name them x1 and x2 .

This formulas both define the same decision boundary:

ys = (-theta[0] - theta[1] * xs) / theta[2]
theta[2] * ys = (-theta[0] - theta[1] * xs)

But to plot the boundary you should define one feature with other.

So here ys are not your predictions your prediction is the sign of this expression which depends of two features xs and ys :

 theta[0] + theta[1] * xs + theta[2] * ys

on the plot this is line which divide your points in two groups. I attached screenshot from your link where this describes.