class OLSRegression:
Regression = linear_model.LinearRegression()
def __init__(self, DataFileName, IndependentVariablesList, DependentVariablesList):
self.Data = DataSet(DataFileName)
self.TrainingData = self.Data.TrainingData()
self.TestingData = self.Data.TestingData()
self.DependentVariablesList = DependentVariablesList
self.IndependentVariablesList = IndependentVariablesList
self.CalculateAICs()
def CalculateAICs(self):
for IndependentVariable in self.IndependentVariablesList:
self.IndependentVariable = IndependentVariable
self.DoTheRegression()
print(self.LogMessage())
def DoTheRegression(self):
AICResetingFactor = 10e10
self.LowestAIC = AICResetingFactor
IteractionsNumber = len(self.DependentVariablesList)
for i in range(IteractionsNumber):
self.FindLowestAICs()
def FindLowestAICs(self):
import random
random.shuffle(self.DependentVariablesList)
self.DependentVariables = []
for Variable in self.DependentVariablesList:
self.DependentVariables.append(Variable)
self.SetLowestAIC()
self.SaveResults()
def SetLowestAIC(self):
self.TrainTheModel()
if(self.AIC() < self.LowestAIC):
self.LowestAIC = self.AIC()
def TrainTheModel(self):
self.Regression = linear_model.LinearRegression()
self.Regression.fit(self.TrainingData[self.DependentVariables], self.TrainingData[self.IndependentVariable])
def SaveResults(self):
Index = self.Data.AllData().index.values
Figure = plt.figure()
SubImage = plt.subplot(111)
SubImage.plot(Index, self.Data.AllData()[self.IndependentVariable], color='blue', label='$y = Data' )
SubImage.plot(Index, self.Regression.predict(self.Data.AllData()[self.DependentVariables]), color='green', label='$y = Model')
FigureName = self.IndependentVariable + ' AIC ' + str(self.LowestAIC)
plt.title(FigureName)
self.LogFolderName = 'Logs/' + self.IndependentVariable + '/'
self.CreateLogFolder()
Figure.savefig(self.LogFolderName + FigureName + '.png')
self.WriteResults()
def CreateLogFolder(self):
if not os.path.exists(self.LogFolderName):
os.makedirs(self.LogFolderName)
def LogMessage(self):
import datetime
Log = '\n' + str(datetime.datetime.now())
Log += '\nLowest AIC for: ' + str(self.IndependentVariable) +' '+ str(self.LowestAIC)+ '\nFound using: ' + str(self.DependentVariables)
Log += '\n' + str(self.Regression.intercept_) + str(self.Regression.coef_)
Log += '\nMSE :' + str(mean_squared_error(self.TestingData[self.IndependentVariable], self.ModelFit()))
Log += '\nAIC: ' + str(self.AIC())
return Log
def AIC(self):
import numpy as np
residuals = self.ModelFit() - self.TestingData[self.IndependentVariable]
NonZeroValue = 0.000000000000000000001
SumOfSquaredErrors = sum(np.power(residuals,2)) + NonZeroValue
NumberOfVariables= len(self.DependentVariables) + 1
NumberOfInstances = len(self.ModelFit())
AIC = (NumberOfInstances * np.log(SumOfSquaredErrors / NumberOfInstances)) + (2 * NumberOfVariables)
return AIC
def WriteResults(self):
import time
timestr = time.strftime("%Y%m%d-%H%M%S")
self.file = open(self.LogFolderName + 'AIC' + str(self.LowestAIC) + timestr + '.log','a')
self.file.write(self.LogMessage())
self.file.close()
def ModelFit(self):
TestingData = self.TestingData[self.DependentVariables]
ModelFit = self.Regression.predict(TestingData)
return ModelFit
def AICPlot(self):
Index = self.Data.AllData().index.values
Figure = plt.figure()
SubImage = plt.subplot(111)
SubImage.plot(Index, self.Data.AllData()[self.IndependentVariable], color='blue', label='$y = Data' )
SubImage.plot(Index, self.Regression.predict(self.Data.AllData()[self.DependentVariables]), color='green', label='$y = Model')
FigureName = self.IndependentVariable + ' AIC ' + str(self.LowestAIC)
plt.title(FigureName)
plt.show()
Figure.savefig('Logs/' + FigureName + '.png')
