LinearRegessorPython.py

class LinearRegressor:
    def __init__(self, iters, X, y, lr):
        self.iters = iters
        self.X = X
        self.lr = lr
        self.y = y
        self.b = 0
        self.w = 1
    def hypothesis_function(self, x):
        list_of_hypothesis_functions = []
        for i in range(len(x)):
            list_of_hypothesis_functions.append(self.b + x[i] * self.w)
        return list_of_hypothesis_functions
    
    def cost_function(self):
        costs = []
        costs2 = []
        for i in range(len(self.X)):
            costs.append(self.hypothesis_function(self.X)[i] - self.y[i])
            costs.append((self.hypothesis_function(self.X)[i] - self.y[i]) * self.X[i])
        return sum(costs) / len(self.X), sum(costs2) / len(self.X)
    
    def gradient_descent(self):
        print(f"previous cost: {self.cost_function()[0]}")
        bn = self.b - self.lr * self.cost_function()[0]
        wn = self.w - self.lr * self.cost_function()[1]
        self.b = bn
        self.w = wn
        return self.b, self.w, self.cost_function()[0]
    
    def train(self):
        for i in range(self.iters + 1):
            bout, wout, cost = self.gradient_descent()
            print(f"bias: {bout}, weight: {wout}, cost: {cost}\n")
    def predict(self, x):
        return self.hypothesis_function(x)

# Example
model = LinearRegressor(iters=2, X=5, y=10, lr=0.01)
model.train()