Chapter 6: u_y unused

[ndrsmchl]

Hi, I noticed that u_y is unused in chapter 6's GPSMemorySCM class. Should it be included in line memory = 0.7*hippocampus + 0.25*u like memory = u_y + 0.7*hippocampus + 0.25*u (or with 0.25*)?

# First, we'll build a structural causal model (SCM)
class GPSMemorySCM:
    
    def __init__(self, random_seed=None):
        self.random_seed = random_seed
        self.u_x = stats.truncnorm(0, np.infty, scale=5)
        self.u_y = stats.norm(scale=2)
        self.u_z = stats.norm(scale=2)
        self.u = stats.truncnorm(0, np.infty, scale=4)
        
    def sample(self, sample_size=100, treatment_value=None):
        """Samples from the SCM"""
        if self.random_seed:
            np.random.seed(self.random_seed)
        
        u_x = self.u_x.rvs(sample_size)
        u_y = self.u_y.rvs(sample_size)
        u_z = self.u_z.rvs(sample_size)
        u = self.u.rvs(sample_size)
        
        if treatment_value:
            gps = np.array([treatment_value]*sample_size)
        else:
            gps = u_x + 0.7*u
            
        hippocampus = -0.6*gps + 0.25*u_z
        memory = 0.7*hippocampus + 0.25*u
        
        return gps, hippocampus, memory
    
    def intervene(self, treatment_value, sample_size=100):
        """Intervenes on the SCM"""
        return self.sample(treatment_value=treatment_value, sample_size=sample_size)

[AlxndrMlk]

Hi @ndrsmchl

Thank you for opening the issue.

You're correct, the u_y term was mistakenly omitted in the code. It should have its own coefficient, for example:

memory = 0.7*hippocampus + 0.25*u + 0.05*u_y

I added it to my notes and we'll fix it for the 2nd edition of the book.

Thank you!