Implement Gradient Descent | Dataford Interview Questions

Plymouth Rock Assurance is prototyping a simple claim-severity model: prediction = weight * x + bias. Implement batch gradient descent to minimize mean squared error across training examples.

Formal Specification

Implement gradient_descent(x, y, learning_rate, iterations).

x is a list of numeric feature values.
y is a list of numeric target values, where y[i] corresponds to x[i].
Start with weight = 0.0 and bias = 0.0.
For each iteration, compute gradients over the entire dataset before updating either parameter:
- error_i = weight * x[i] + bias - y[i]
- d_weight = (2 / n) * sum(error_i * x[i])
- d_bias = (2 / n) * sum(error_i)
Update both parameters using the learning rate.
Return [weight, bias], rounded to 10 decimal places.

Constraints

1 <= len(x) == len(y) <= 10^4
-10^3 <= x[i], y[i] <= 10^3
0 < learning_rate <= 1
0 <= iterations <= 10^4
Return weight and bias rounded to 10 decimal places

Function Signature

def gradient_descent(x, y, learning_rate, iterations):

Your question is Implement Gradient Descent. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.

Problem