Estadísticas y predicciones de Second League Division B Group 1 Relegation Group

Estatísticas Clave

• Goles anotados por equipo: Analizar cuántos goles ha marcado cada equipo puede darte una idea de su potencial ofensivo.
• Goles recibidos por equipo: Esto te ayudará a entender qué tan sólida es la defensa del equipo contrario.
• Promedio de posesión: Un alto porcentaje de posesión suele indicar control del juego y dominio territorial.
• Tiros a puerta: Cuantos más tiros a puerta tenga un equipo, mayores son sus posibilidades de marcar goles.

Tácticas Comunes

• Juego directo: Algunos equipos prefieren pasar directamente al ataque sin elaborar demasiado el juego desde atrás. Esto puede ser efectivo contra defensas estáticas.
# -*- coding: utf-8 -*- """ Created on Thu May 28 13:18:44 2020 @author: KUBICZ Michał """ import os import sys import numpy as np import pandas as pd def load_data(path): # read data from csv file df = pd.read_csv(path) # define labels df['label'] = df['TARGET'].replace({1:'no default', 0:'default'}) # drop unused columns df.drop(columns=['ID', 'TARGET'], inplace=True) return df def preprocess_data(df): # check if there are any null values in the dataset if df.isnull().values.any(): # calculate the mean for each column with null values and fill it with it's mean for column in df.columns: if df[column].isnull().values.any(): df[column].fillna(df[column].mean(), inplace=True) # remove outliers using IQR method Q1 = df.quantile(0.25) Q3 = df.quantile(0.75) IQR = Q3 - Q1 df = df[~((df<(Q1 - 1.5 * IQR)) | (df > (Q3 + 1.5 * IQR))).any(axis=1)] return df def split_data(df): # split data into features and labels X = df.drop('label', axis=1).values Y = df['label'].values return X,Y def balance_data(X,Y): # count number of each label in the dataset counts = np.unique(Y, return_counts=True)[1] count_default = counts[0] count_no_default = counts[1] # calculate number of elements to be added to the default class add_count_default = count_no_default - count_default # get indices of samples with default label indices_default = np.where(Y == 'default')[0] # add samples from the default class to the training set by random choice with replacement X_balanced = np.concatenate((X,Y), axis=0) X_balanced = np.append(X_balanced,np.random.choice(X_balanced[indices_default], size=add_count_default), axis=0) Y_balanced = np.concatenate((Y,Y), axis=0) Y_balanced = np.append(Y_balanced,np.random.choice(Y_balanced[indices_default], size=add_count_default), axis=0) return X_balanced,Y_balanced def one_hot_encode(Y): # convert labels to one hot encoding Y_one_hot = [] for label in Y: if label == 'no default': Y_one_hot.append([1.,0.]) elif label == 'default': Y_one_hot.append([0.,1.]) Y_one_hot = np.array(Y_one_hot) return Y_one_hot<|repo_name|>KubiczMichal/Default-prediction-using-neural-networks<|file_sep|>/README.md # Default-prediction-using-neural-networks<|file_sep|># -*- coding: utf-8 -*- """ Created on Wed May 27 22:17:51 2020 @author: KUBICZ Michał """ import os import sys import numpy as np from tensorflow.keras.models import Sequential from tensorflow.keras.layers import Dense from sklearn.model_selection import train_test_split from helper_functions import * # path to data directory path_to_data_directory = os.path.join(os.getcwd(), 'data') # path to data file path_to_data_file = os.path.join(path_to_data_directory,'data.csv') # load data from csv file df_raw_data = load_data(path_to_data_file) # preprocess raw data df_preprocessed_data = preprocess_data(df_raw_data) # split preprocessed data into features and labels X,Y = split_data(df_preprocessed_data) # balance the data by adding more samples to the minority class X_balanced,Y_balanced = balance_data(X,Y) # one hot encode labels Y_one_hot_encoded = one_hot_encode(Y_balanced) # split balanced and preprocessed data into training and testing sets X_train,X_test,Y_train,Y_test = train_test_split(X_balanced,Y_one_hot_encoded,test_size=0.33) # define model architecture model = Sequential() model.add(Dense(units=32,input_dim=24,kernel_initializer='normal',activation='relu')) model.add(Dense(units=16,kernel_initializer='normal',activation='relu')) model.add(Dense(units=8,kernel_initializer='normal',activation='relu')) model.add(Dense(units=2,kernel_initializer='normal',activation='softmax')) # compile model model.compile(loss='categorical_crossentropy',optimizer='adam',metrics=['accuracy']) # fit model to training data history_object=model.fit(X_train,Y_train,batch_size=10000,nb_epoch=10000, validation_split=0.33,callbacks=[EarlyStopping(monitor='val_loss', patience=10, verbose=0, mode='auto')]) # evaluate model on test data scores=model.evaluate(X_test,Y_test,batch_size=10000) print('Loss:',scores[0]) print('Accuracy:',scores[1]) # plot model accuracy and loss during training plt.plot(history_object.history['accuracy']) plt.plot(history_object.history['val_accuracy']) plt.title('Model Accuracy') plt.ylabel('Accuracy') plt.xlabel('Epoch') plt.legend(['Train','Validation'], loc='upper left') plt.show() plt.plot(history_object.history['loss']) plt.plot(history_object.history['val_loss']) plt.title('Model Loss') plt.ylabel('Loss') plt.xlabel('Epoch') plt.legend(['Train','Validation'], loc='upper left') plt.show()<|file_sep|># -*- coding: utf-8 -*- """ Created on Fri Jun 12 22:14:49 2020 @author: KUBICZ Michał """ import numpy as np from tensorflow.keras.models import Sequential from tensorflow.keras.layers import Dense from sklearn.model_selection import train_test_split from helper_functions import * path_to_data_directory=os.path.join(os.getcwd(),'data') path_to_data_file=os.path.join(path_to_data_directory,'data.csv') df_raw_data=load_data(path_to_data_file) df_preprocessed_data=preprocess_data(df_raw_data) X,Y=split_data(df_preprocessed_data) X_balanced,Y_balanced=balance_data(X,Y) Y_one_hot_encoded=one_hot_encode(Y_balanced) X_train,X_test,Y_train,Y_test=train_test_split(X_balanced,Y_one_hot_encoded,test_size=.33) model=Sequential() model.add(Dense(units=32,input_dim=24,kernel_initializer='normal',activation='relu')) model.add(Dense(units=16,kernel_initializer='normal',activation='relu')) model.add(Dense(units=8,kernel_initializer='normal',activation='relu')) model.add(Dense(units=2,kernel_initializer='normal',activation='softmax')) model.compile(loss='categorical_crossentropy',optimizer='adam',metrics=['accuracy']) history_object=model.fit(X_train,Y_train,batch_size=10000,nb_epoch=10000, validation_split=.33,callbacks=[EarlyStopping(monitor='val_loss', patience=10, verbose=False, mode='auto')]) scores=model.evaluate(X_test,Y_test,batch_size=10000) print('Loss:',scores[0]) print('Accuracy:',scores[1]) predictions=model.predict_classes(X_test,batch_size=10000) correct_predictions=[] incorrect_predictions=[] for i in range(len(predictions)): if predictions[i] == Y_test[i]: correct_predictions.append(i) else: incorrect_predictions.append(i)<|repo_name|>SebastianSzczygiel/AutoML<|file_sep|>/src/data/DataUtils.pyx # cython: language_level=3 cimport cython @cython.boundscheck(False) # Deactivate bounds checking for entire function. @cython.wraparound(False) # Deactivate negative indexing for entire function. def find_elements(np.ndarray[np.float64_t] arr, double value): cdef int i,j,n,m n = arr.shape[0] m = len(value) cdef list result=[] for i in range(n): for j in range(m): if arr[i] == value[j]: result.append(i) break @cython.boundscheck(False) # Deactivate bounds checking for entire function. @cython.wraparound(False) # Deactivate negative indexing for entire function. def find_elements_by_range(np.ndarray[np.float64_t] arr, double value_1, double value_2):