amaarora

class TreeEnsemble():
    def __init__(self, x, y, n_trees, sample_sz, min_leaf=5):
        np.random.seed(42)
        self.x,self.y,self.sample_sz,self.min_leaf = x,y,sample_sz,min_leaf
        self.trees = [self.create_tree() for i in range(n_trees)]

    def create_tree(self):
        rnd_idxs = np.random.permutation(len(self.y))[:self.sample_sz]
        return DecisionTree(self.x.iloc[rnd_idxs], self.y[rnd_idxs], min_leaf=self.min_leaf)
        

amaarora

class DecisionTree():
    def __init__(self, x, y, idxs=None, min_leaf=5):
        if idxs is None: idxs=np.arange(len(y))
        self.x,self.y,self.idxs,self.min_leaf = x,y,idxs,min_leaf
        self.n,self.c = len(idxs), x.shape[1]
        self.val = np.mean(y[idxs])
        self.score = float('inf')
        self.find_varsplit()
        
    # This just does one decision; we'll make it recursive later

amaarora

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

    for i in range(self.n):
        lhs = x<=x[i]
        rhs = x>x[i]
        if rhs.sum()<self.min_leaf or lhs.sum()<self.min_leaf: continue
        lhs_std = y[lhs].std()
        rhs_std = y[rhs].std()
        curr_score = lhs_std*lhs.sum() + rhs_std*rhs.sum()

amaarora

def find_varsplit(self):
    for i in range(self.c): self.find_better_split(i)
    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 = DecisionTree(self.x, self.y, self.idxs[lhs])
    self.rhs = DecisionTree(self.x, self.y, self.idxs[rhs])

amaarora

class TreeEnsemble():
    def __init__(self, x, y, n_trees, sample_sz, min_leaf=5):
        np.random.seed(42)
        self.x,self.y,self.sample_sz,self.min_leaf = x,y,sample_sz,min_leaf
        self.trees = [self.create_tree() for i in range(n_trees)]

    def create_tree(self):
        idxs = np.random.permutation(len(self.y))[:self.sample_sz]
        return DecisionTree(self.x.iloc[idxs], self.y[idxs], 
                    idxs=np.array(range(self.sample_sz)), min_leaf=self.min_leaf)

amaarora

# coding: utf-8

# In[1]:


import numpy as np
import pandas as pd
from matplotlib import pyplot as plt
from sklearn.ensemble import RandomForestRegressor

amaarora

def get_weights(*dims): return nn.Parameter(torch.randn(dims)/dims[0])
def softmax(x): return torch.exp(x)/(torch.exp(x).sum(dim=1)[:,None])

class LogReg(nn.Module):
    def __init__(self):
        super().__init__()
        self.l1_w = get_weights(28*28, 10)  # Layer 1 weights
        self.l1_b = get_weights(10)         # Layer 1 bias

    def forward(self, x):

amaarora

%%time
scores={}
for col in X_valid.columns:
    X = X_valid.copy()
    X[col] = np.random.choice(X[col], len(X))
    scores[col]=rmse(m.predict(X), y_valid)-base

amaarora

def reward_function(params):
    '''
   Example of rewarding the agent to stay within boundary
    '''
    import math
    # Read input parameters
    track_width = params['track_width']
    distance_from_center = params['distance_from_center']
    steering = abs(params['steering_angle'])
    all_wheels_on_track = params['all_wheels_on_track']

amaarora

def reward_function(params):
    '''
    Example of rewarding the agent to follow center line
    '''
    import math
    # Read input parameters
    track_width = params['track_width']
    distance_from_center = params['distance_from_center']
    steering = abs(params['steering_angle'])
    all_wheels_on_track = params['all_wheels_on_track']