EM One M-Step Lite

Problem

Perform one lite EM M-step for the observation noise variance: r_new = mean((x - level)^2) + p, where p is the state covariance from the E-step.

Return p when observations is empty.

Hint

The state uncertainty p inflates the raw residual variance.

case 1

in: [0,1,[1,2,3]]

expected: 5.666666666666667

case 2

in: [1,0,[1,1,1]]

expected: 0

case 3

in: [0,2,[]]

expected: 2

case 4

in: [2,0.5,[1,3]]

expected: 1.5

Starter codePython

def em_one_m_step(level, p, observations):
    # Your code here
    pass

4 test cases · run your code against them in the editor