Eligijus112

# Data wrangling
import pandas as pd 

# Infinity constant 
from math import inf 


class Tree():
    """
    Class to fit a regression tree to the given data 

Eligijus112

# The base class with the weak learner 
from regression.tree import Tree 

# Data wrangling 
import pandas as pd 

# Python infinity
from math import inf

Eligijus112

# Dataset loading 
from sklearn.datasets import fetch_california_housing

# Loading the data
_cali_data = fetch_california_housing(as_frame=True)

# Features and target
X, y = _cali_data.data, _cali_data.target

# Droping the geo coordinate featuress

Eligijus112

# Train test spliting 
from sklearn.model_selection import train_test_split

# Importing the sklearn implementation 
from sklearn.tree import DecisionTreeRegressor

# Precision metrics 
from sklearn.metrics import mean_absolute_error

# Spliting the data into training and testing sets

Eligijus112

# Entries in the nodes 
n_entries = {
    "node 1": 15480,
    "node 2": 12163,
    "node 3": 3317,
    "node 4": 5869,
    "node 5": 6294,
    "node 6": 2317,
    "node 7": 1000, 
    "node 8": 2454,

Eligijus112

def node_importance(
    node_main: str, 
    node_left: str, 
    node_right: str, 
    n_entries_weight: dict, 
    i_sq: dict
    ) -> float:
    """
    Calculated the importance of the node_main
    

Eligijus112

# Calculating the importance of each node
importance = {}
for node in node_pairs:
    importance[node] = round(node_importance(node, *node_pairs[node], n_entries_weight, i_sq), 3)

print(f"Node importance: {importance}")

# Going from node importance to feature importance
feature_importance = {}
for node in node_pairs: 

Eligijus112

# Loading the data
_cali_data = fetch_california_housing(as_frame=True)
X, y = _cali_data.data, _cali_data.target

# Droping the geo coordinate featuress
X = X.drop(columns=['Latitude', 'Longitude'])

# Droping the population feature; In real life modeling, this could be used as weight. 
# For educational and inference purposes, we drop it.
X = X.drop(columns=['Population'])

Eligijus112

# Importing the package
import tensorflow as tf


class NMSE(tf.keras.losses.Loss):
    def __init__(self):
        super().__init__()

    def call(self, y_true, y_pred):
        # Calculating the mse; 

Eligijus112

def elastic_net_to_keras(alpha: float, l1_ratio: float):
    """
    Converts ElasticNet parameters from sklearn to Keras regularizers.
    
    Arguments
    ---------
    alpha: float
        The regularization strength of the model.
    l1_ratio: float
        The l1 regularization ratio of the model.