def test_pais_100(N, indicators):
g08.testing = True
g08.riggedRandom = 100
muestra = g08.generar_muestra_pais(N)
for i in range(len(indicators)):
assert (g08.show_percentages_indicator(muestra, indicators[i]) == 1.0)
def test_pais_90(N, indicators, r):
g08.testing = True
g08.riggedRandom = 90
muestra = g08.generar_muestra_pais(N)
for i in range(len(indicators)):
assert (0.9 - r < g08.show_percentages_indicator(
muestra, indicators[i]) < 0.9 + r)
def averge(n): corridas_pais = [] result = [] keys = [] for i in range(10): if actual!="Pais": keys, values = show_percentages(generar_muestra_provincia(n, actual.upper())) elif actual=="Pais": keys, values = show_percentages(generar_muestra_pais(n)) corridas_pais.append(values) corridas_pais = average_aux(corridas_pais) return keys, corridas_pais
def gen_dataset(n, provincia, sample_type=1):
"""
Generates the working dataset
"""
result = None
if provincia != "PAIS":
result = g08.generar_muestra_provincia(n,provincia,sample_type)
else:
result = g08.generar_muestra_pais(n,sample_type)
return result
def test_distribucion_por_provincia(N):
dicc = {
"SAN JOSE": 0,
"ALAJUELA": 1,
"CARTAGO": 2,
"HEREDIA": 3,
"GUANACASTE": 4,
"PUNTARENAS": 5,
"LIMON": 6
}
muestraPop = [0, 0, 0, 0, 0, 0, 0]
population = [
1404242, #SJO
848146, #ALA
490903, #CAR
433677, #HER
326953, #GUA
410929, #PUN
386862
] #LIM
totalPop = sum(population)
distribucionPProvincia = [0, 0, 0, 0, 0, 0, 0]
distribucionPProvinciaMuestra = [0, 0, 0, 0, 0, 0, 0]
for i in range(len(population)):
distribucionPProvincia[i] = population[i] / totalPop
muestra1 = g08.generar_muestra_pais(N)
for i in range(len(muestra1)):
muestraPop[dicc[muestra1[i][32]]] += 1
for i in range(len(muestraPop)):
distribucionPProvinciaMuestra[i] = muestraPop[i] / N
for i in range(len(population)):
assert (distribucionPProvincia[i] - 0.01 <
distribucionPProvinciaMuestra[i] <
distribucionPProvincia[i] + 0.01)
import pandas from tec.ic.ia.pc1 import g08 from sklearn.preprocessing import MinMaxScaler from keras.models import Sequential from keras.layers import Dense, Dropout from keras.wrappers.scikit_learn import KerasClassifier from keras.utils import np_utils from sklearn.cross_validation import train_test_split from sklearn.preprocessing import LabelEncoder, StandardScaler # fix random seed for reproducibility seed = 5 numpy.random.seed(seed) # load dataset dataframe = numpy.array(g08.generar_muestra_pais(10000, 0)) X = dataframe[:, 1:-2] scalerX = StandardScaler() scalerX.fit(X) X = scalerX.transform(X) Y = dataframe[:, -1] # encode class values as integers encoder = LabelEncoder() encoder.fit(Y) encoded_Y = encoder.transform(Y) # convert integers to dummy variables (i.e. one hot encoded) dummy_y = np_utils.to_categorical(encoded_Y)
idxstr = 'predictions_' + str(round_number)
predictions[idxstr] += prediction
idxstr = 'accuracies_' + str(round_number)
accuracies[idxstr].append(accuracy)
print("Accuracy: ", accuracy)
is_training = ['False' for i in range(len(training_dataset))]
best_trees = get_best_trees(trees, accuracies)
avg_train_acc = (sum(accuracies['accuracies_1']) + sum(
accuracies['accuracies_2']) + sum(accuracies['accuracies_3'])) / 3
final_dict = final_tests(best_trees, test_dataset, predictions,
avg_train_acc, is_training)
"""
print("Results")
print(len(final_dict['res_1']))
print(len(final_dict['res_2']))
print(len(final_dict['res_3']))
print(len(final_dict['train_set']))
print(final_dict['err_train'])
print(final_dict['err_test'])
"""
return final_dict
if __name__ == '__main__':
cross_validate(dataset=g08.generar_muestra_pais(2005, 1))
third = [g08.PARTIDOS2[int(predictions[i])] for i in range(len(predictions))]
predictions = model.predict(x_test)
third += [g08.PARTIDOS2[int(predictions[i])] for i in range(len(predictions))]
third_acc_train = model.score(x_train,y_train.ravel())
third_acc = model.score(x_test,y_test.ravel())
#print(first)
print(first_acc)
print()
#print(second)
print(second_acc)
print()
#print(third)
print(third_acc)
finalDict = {
'res_1': first,
'res_2': second,
'res_3': third,
'err_train': (first_acc+second_acc+third_acc)/3,
'err_test': (first_acc_train+second_acc_train+third_acc_train)/3,
'train_set': [True]*len(X1)+[False]*len(Y1)
}
return finalDict
execute_model(g08.generar_muestra_pais(10000,1),20)
predict = max(set(bestOccurrences), key=bestOccurrences.count)
if predict == testPerson[-1]:
accuracy += 1
finalDict[resString].append(g08.PARTIDOS[int(predict)])
accuracy = accuracy / len(dataSet)
totalaccuracy += accuracy
dataSetIndex += 1
print("Error ", dataSetIndex, ": ", 1 - accuracy)
finalDict['err_test'] = totalaccuracy / 3
return finalDict
# Returns the best trees for each round based on the precision they had
def getBestTrees(trees1, trees2, trees3, accuracies1, accuracies2,
accuracies3):
print("Best accuracy 1", max(accuracies1))
ind1 = accuracies1.index(max(accuracies1))
print("Best accuracy 2", max(accuracies2))
ind2 = accuracies2.index(max(accuracies2))
print("Best accuracy 3", max(accuracies3))
ind3 = accuracies3.index(max(accuracies3))
return trees1[ind1], trees2[ind2], trees3[ind3]
if __name__ == '__main__':
cross_validate(dataset=numpy.array(g08.generar_muestra_pais(1003, 1)))