Головна » Basketball » Sweden » Basketligan

Basketligan stats & predictions

Yesterday

Tomorrow

">

Preguntas Frecuentes sobre Apuestas en Baloncesto

<|repo_name|>taijihuang/nn<|file_sep|>/code/NN4.py # -*- coding: utf-8 -*- """ Created on Wed Mar 25 10:52:41 2020 @author: Administrator """ import numpy as np import pandas as pd import os from sklearn.preprocessing import MinMaxScaler import matplotlib.pyplot as plt # os.chdir('C:/Users/Administrator/Desktop/深度学习实验/实验三') # file = 'nn4.csv' # # file = 'nn4.csv' # data = pd.read_csv(file) # data = data.iloc[:,1:] # # scaler = MinMaxScaler() # data_scale = scaler.fit_transform(data) # # # 训练集、测试集 # train_data = data_scale[:int(len(data_scale)*0.7), :] # test_data = data_scale[int(len(data_scale)*0.7):, :] # # x_train = train_data[:,:-1] # y_train = train_data[:,-1:] # # x_test = test_data[:,:-1] # y_test = test_data[:,-1:] x_train = np.array([[0,0],[0,1],[1,0],[1,1]]) y_train = np.array([[0],[1],[1],[0]]) x_test = np.array([[0,0],[0,1],[1,0],[1,1]]) y_test = np.array([[0],[1],[1],[0]]) np.random.seed(3) class NN(): def __init__(self): self.W_01 = np.random.rand(3,2) self.W_12 = np.random.rand(3) self.b_01 = np.random.rand(3) self.b_12 = np.random.rand(3) self.learning_rate_01 = np.ones((3,2)) self.learning_rate_12 = np.ones((3)) self.dw_01 = np.zeros((3,2)) self.dw_12 = np.zeros((3)) self.db_01 = np.zeros((3)) self.db_12 = np.zeros((3)) self.dW_01_history_list=[] self.dW_12_history_list=[] self.db_01_history_list=[] self.db_12_history_list=[] self.loss_history_list=[] def sigmoid(self,x): return (1/(1+np.exp(-x))) def sigmoid_prime(self,x): return x*(1-x) nn=NN() EPOCHS=10000 for epoch in range(EPOCHS): for i in range(len(x_train)): #前向传播 #反向传播 #更新参数 #记录损失 print('Epoch:',epoch,'loss:',loss)<|file_sep|># -*- coding: utf-8 -*- """ Created on Mon Mar 23 21:09:45 2020 @author: Administrator """ import numpy as np import pandas as pd import os from sklearn.preprocessing import MinMaxScaler import matplotlib.pyplot as plt os.chdir('C:/Users/Administrator/Desktop/深度学习实验/实验三') file='nn3.csv' data=pd.read_csv(file) scaler=MinMaxScaler() data_scale=scaler.fit_transform(data) train_data=data_scale[:int(len(data_scale)*0.7),:] test_data=data_scale[int(len(data_scale)*0.7):,:] x_train=train_data[:,:-1] y_train=train_data[:,-1:] x_test=test_data[:,:-1] y_test=test_data[:,-1:] np.random.seed(3) class NN(): def __init__(self): self.W=np.random.rand(6,5) self.b=np.random.rand(6) self.learning_rate=np.ones((6,)) self.dw=np.zeros((6,)) self.db=np.zeros((6,)) self.loss_history_list=[] nn=NN() EPOCHS=10000 for epoch in range(EPOCHS): for i in range(len(x_train)): #前向传播 #反向传播 #更新参数 #记录损失 print('Epoch:',epoch,'loss:',loss)<|file_sep|># -*- coding: utf-8 -*- """ Created on Wed Mar 25 10:52:41 2020 @author: Administrator """ import numpy as np import pandas as pd import os from sklearn.preprocessing import MinMaxScaler import matplotlib.pyplot as plt os.chdir('C:/Users/Administrator/Desktop/深度学习实验/实验三') file='nn4.csv' data=pd.read_csv(file) scaler=MinMaxScaler() data_scale=scaler.fit_transform(data) train_data=data_scale[:int(len(data_scale)*0.7),:] test_data=data_scale[int(len(data_scale)*0.7):,:] x_train=train_data[:,:-1] y_train=train_data[:,-1:] x_test=test_data[:,:-1] y_test=test_data[:,-1:] np.random.seed(3) class NN(): def __init__(self): self.W_01=np.random.rand(3,2) self.W_12=np.random.rand(3) self.b_01=np.random.rand(3) self.b_12=np.random.rand(3) self.learning_rate_01=np.ones((3,2)) self.learning_rate_12=np.ones((3)) self.dw_01=np.zeros((3,2)) self.dw_12=np.zeros((3)) self.db_01=np.zeros((3)) self.db_12=np.zeros((3)) #初始化学习率变化记录列表 nn=NN() EPOCHS=10000 for epoch in range(EPOCHS): for i in range(len(x_train)): z_h=self.sigmoid(np.dot(x_train[i],self.W_01)+self.b_01) z_o=self.sigmoid(np.dot(z_h,self.W_12)+self.b_12) loss=(z_o-y_train[i])**2 dz_o=z_o*(1-z_o)*(z_o-y_train[i]) dz_h=z_h*(1-z_h)*np.dot(dz_o,self.W_12.T) dw_o=dz_o*z_h dw_h=dz_h*x_train[i] db_o=dz_o db_h=dz_h gradient_W_o=self.learning_rate_12*self.dw_o+self.learning_rate_decay*dw_o gradient_W_h=self.learning_rate_01*self.dw_h+self.learning_rate_decay*dw_h gradient_b_o=self.learning_rate_b*self.db_o+self.learning_rate_decay*db_o gradient_b_h=self.learning_rate_b*self.db_h+self.learning_rate_decay*db_h gradient_W_o=(gradient_W_o+gradient_W_o)/np.linalg.norm(gradient_W_o)+epsilon*np.sign(self.W_O) gradient_W_h=(gradient_W_h+gradient_W_h)/np.linalg.norm(gradient_W_h)+epsilon*np.sign(self.W_H) gradient_b_o=(gradient_b_o+gradient_b_o)/np.linalg.norm(gradient_b_o)+epsilon*np.sign(self.b_O) gradient_b_h=(gradient_b_h+gradient_b_h)/np.linalg.norm(gradient_b_h)+epsilon*np.sign(self.b_H) <|file_sep|># -*- coding: utf-8 -*- """ Created on Mon Mar 23 21:09:45 2020 @author: Administrator """ import numpy as np import pandas as pd import os from sklearn.preprocessing import MinMaxScaler import matplotlib.pyplot as plt os.chdir('C:/Users/Administrator/Desktop/深度学习实验/实验三') file='nn5.csv' data=pd.read_csv(file) scaler=MinMaxScaler() data_scale=scaler.fit_transform(data) train_data=data_scale[:int(len(data_scale)*0.7),:] test_data=data_scale[int(len(data_scale)*0.7):,:] x_train=train_data[:,:-1] y_train=train_data[:,-1:] x_test=test_data[:,:-1] y_test=test_data[:,-1:] np.random.seed(3) class NN(): def __init__(self): #神经网络初始化 nn=NN() EPOCHS=10000 for epoch in range(EPOCHS): for i in range(len(x_train)): <|repo_name|>alexskeels/bc2019<|file_sep|>/README.md ## BC2019 Analysis ### Data download and preparation Download the data from the [BC2019 website](https://bc2019.ca/en/data). This is a zip file containing all of the results (in CSV format) and metadata (in Excel format). The files are named `raw_{filename}.csv` and `raw_{filename}.xlsx`, respectively. In the `prepare.py` script there are two functions: * `parse_metadata()`: this reads the Excel files and creates one Python dictionary per Excel file. * `parse_results()`: this reads the CSV files and creates one Pandas DataFrame per CSV file. The dictionary keys are the names of the contests and each value is another dictionary with the following keys: * `name`: contest name. * `level`: level of government (federal or provincial). * `date`: election date. * `num_seats`: number of seats. * `seat_type`: type of seat (single-member or multi-member). * `voter_turnout`: voter turnout (percentage). * `results`: DataFrame with results. * `parties`: list of parties that ran candidates. * `candidates`: list of candidates that ran. The DataFrame keys are the names of the contests and each value is another DataFrame with columns: * rid (or rid/municipality): riding or municipality name. * candidate_id: candidate identifier. * candidate_name: candidate name. * party_id: party identifier. * party_name: party name. * votes_cast_for_candidate: number of votes cast for the candidate. * total_valid_votes_cast_in_riding_or_municipality: total number of valid votes cast in riding or municipality. * percentage_of_valid_votes_cast_for_candidate_in_riding_or_municipality: percentage of valid votes cast for the candidate in riding or municipality. ### Data analysis The file `analysis.py` contains some code to analyze the data. The code can be run by calling one of these commands: bash $ python analysis.py -a rank_by_seat -o rank_by_seat.pdf --date "May 2019" $ python analysis.py -a rank_by_voters -o rank_by_voters.pdf --date "May 2019" $ python analysis.py -a rank_by_voters_and_seat -o rank_by_voters_and_seat