curiousily

class DecisionTreeRegressor:
  
  def fit(self, X, y, min_leaf = 5):
    self.dtree = Node(X, y, np.array(np.arange(len(y))), min_leaf)
    return self
  
  def predict(self, X):
    return self.dtree.predict(X.values)

curiousily

class Node:

    def __init__(self, x, y, idxs, min_leaf=5):
        self.x = x 
        self.y = y
        self.idxs = idxs 
        self.min_leaf = min_leaf
        self.row_count = len(idxs)
        self.col_count = x.shape[1]
        self.val = np.mean(y[idxs])

curiousily

class Node:

    def __init__(self, x, y, idxs, min_leaf=5):
        self.x = x 
        self.y = y
        self.idxs = idxs 
        self.min_leaf = min_leaf
        self.row_count = len(idxs)
        self.col_count = x.shape[1]
        self.val = np.mean(y[idxs])

curiousily

def find_varsplit(self):
    for c in range(self.col_count): self.find_better_split(c)
    if self.is_leaf: return
    x = self.split_col
    lhs = np.nonzero(x <= self.split)[0]
    rhs = np.nonzero(x > self.split)[0]
    self.lhs = Node(self.x, self.y, self.idxs[lhs], self.min_leaf)
    self.rhs = Node(self.x, self.y, self.idxs[rhs], self.min_leaf)

curiousily

@property
def split_col(self): return self.x.values[self.idxs,self.var_idx]

@property
def is_leaf(self): return self.score == float('inf')     

curiousily

def find_better_split(self, var_idx):
    x = self.x.values[self.idxs, var_idx]

    for r in range(self.row_count):
        lhs = x <= x[r]
        rhs = x > x[r]
        if rhs.sum() < self.min_leaf or lhs.sum() < self.min_leaf: continue

        curr_score = self.find_score(lhs, rhs)
        if curr_score < self.score: 

curiousily

def predict(self, x):
    return np.array([self.predict_row(xi) for xi in x])

def predict_row(self, xi):
    if self.is_leaf: return self.val
    node = self.lhs if xi[self.var_idx] <= self.split else self.rhs
    return node.predict_row(xi)

curiousily

regressor = DecisionTreeRegressor().fit(X, y)
preds = regressor.predict(X)

curiousily

X_test = df_test[['OverallQual', 'GrLivArea', 'GarageCars']]
pred_test = regressor.predict(X_test)

submission = pd.DataFrame({'Id': df_test.Id, 'SalePrice': pred_test})
submission.to_csv('submission.csv', index=False)

curiousily

!pip install tensorflow-gpu==2.0.0-alpha0