curiousily

def loss(h, y):
  sq_error = (h - y)**2
  n = len(y)
  return 1.0 / (2*n) * sq_error.sum()

curiousily

x = df_train['GrLivArea']
y = df_train['SalePrice']

x = (x - x.mean()) / x.std()
x = np.c_[np.ones(x.shape[0]), x] 

curiousily

class LinearRegression:
  
  def predict(self, X):
    return np.dot(X, self._W)
  
  def _gradient_descent_step(self, X, targets, lr):

    predictions = self.predict(X)
    
    error = predictions - targets

curiousily

class TestLinearRegression(unittest.TestCase):

    def test_find_coefficients(self):
      clf = LinearRegression()
      clf.fit(x, y, n_iter=2000, lr=0.01)
      np.testing.assert_array_almost_equal(clf._W, np.array([180921.19555322,  56294.90199925]))

curiousily

clf = LinearRegression()
clf.fit(x, y, n_iter=2000, lr=0.01)

curiousily

x = df_train[['OverallQual', 'GrLivArea', 'GarageCars']]

x = (x - x.mean()) / x.std()
x = np.c_[np.ones(x.shape[0]), x] 

curiousily

X = df_train[['OverallQual', 'GrLivArea', 'GarageCars']]
y = df_train['SalePrice']

curiousily

from sklearn.metrics import mean_squared_error
from math import sqrt

def rmse(h, y):
  return sqrt(mean_squared_error(h, y))

curiousily

from sklearn.ensemble import RandomForestRegressor

reg = RandomForestRegressor(
  n_estimators=1, 
  max_depth=2, 
  bootstrap=False, 
  random_state=RANDOM_SEED
)
reg.fit(X, y)

curiousily

preds = reg.predict(X)
metrics.r2_score(y, preds)