breeko

date,confidence
2013-3-26,59.7
2013-2-26,69
2013-1-29,58.6
2012-12-27,65.1
2012-11-27,73.7
2012-10-1,72.2
2012-9-25,70.3
2012-8-28,60.6
2012-7-31,65.9

breeko

DATE,Ratio
11/1/2006,0.91
11/2/2006,0.9
11/3/2006,0.91
11/6/2006,0.76
11/7/2006,0.79
11/8/2006,0.78
11/9/2006,0.84
11/10/2006,1.06
11/13/2006,0.8

breeko

import numpy as np
import gzip
import json

def get_relative_frame(frame, position, new_position=(0,0)):
    """ Rolls a frame such that position is moved to new_position
    Input:
        frame: 2-dimensional array
        position: tuple (row, col)

breeko

import os
import pickle
import tensorflow as tf
from sklearn.model_selection import train_test_split

with open('djma_v3.pkl', 'rb') as input:
    data = pickle.load(input) 
    X = data["X"]
    y = data["y"]

breeko

from keras.models import Sequential
from keras.layers import Dense
from keras.callbacks import EarlyStopping

import pickle
import numpy as np
import json

def get_relative_frame(frame, position, new_position=(0,0)):

breeko

import numpy as np
import gym
from itertools import product

def init_centers(num_splits=4, num_obs=2):
    """ Returns [num_splits**num_obs, num_obs] matrix of equidistant centers from (0,0) to (1,1) """
    return np.array(list(product(np.linspace(0,1,num_splits),repeat=num_obs)))

def init_theta(num_splits=4,num_actions=2):
    """ Returns random [num_splits*num_splits,action_space] matrix of value action pairs """

breeko

# Policy based reinforcement learning agent used to solve openai's CartPole challenge
# https://gym.openai.com/evaluations/eval_dMY1xQiST7GXe4Br5n31w

import numpy as np
import tensorflow as tf
import gym

ENVIRONMENT = "CartPole-v0"
SEED = 0

breeko

import numpy as np
import gym

def evolve(W, sigma, prob_mutate=0.5):
    rand = (np.random.randn(*W.shape) - 0.5) * sigma
    mutate = np.random.choice([0,1],size=W.shape,p=(1-prob_mutate, prob_mutate))
    return W + (mutate * rand)

def evaluate(W, num_trials=1, max_t=1000, render=False):
    cum_rewards = []

breeko

import gym
import numpy as np
from sklearn.preprocessing import StandardScaler
from keras.models import Sequential
from keras.layers import Dense, Dropout
from keras.optimizers import Adamax

# TD Learning

MEMORY = 1                      # Number of prior states to consider when training our agent

breeko

class TicTacToe:
    """ Tic-Tac Toe Board 
        
        Board is represented through a board object
        
        1 ¦ 2 ¦ 3
        --+---+--
        4 ¦ 5 ¦ 6
        --+---+--
        7 ¦ 8 ¦ 9