def ModelIt(name, cut, df, myconfig):
config = ConfigParser.RawConfigParser()
config.read(myconfig)
#MySQL info:
db_username = config.get('DB', 'username')
db_pwd = config.get('DB', 'pwd')
#Database connection:
con = mdb.connect('localhost', db_username, db_pwd, 'InsightPaths')
path = "/home/ubuntu/WebApp/Web/FanGuardFlask/files"
#Prefilter Models:
vp1, vp2, pcutval = GetPFModel(name, con, path)
c0 = GetPred(df, vp1, vp2)
#p0 = pmodel.predict_proba(Xp0)[:,1]
#Spoiler Filter Models:
smodel, vs1, vs2, scutval = GetSFModel(name, con, path, cut)
p1 = modelmaker.model_tester(df, smodel, vs1, vs2)
#c0 = (p0>pcutval).astype(int)
#c0 = c0.reshape(c0.shape[0],1)
c1 = (p1 > scutval).astype(int)
c1 = c1.reshape(c1.shape[0], 1)
return c0 * c1
def Train_A_Model_Direct(tag, model, vocab1,vocab2,df_Train,df_Test):
#Train the actual model:
trained_model,trained_vocab,tagged_vocab = modelmaker.model_trainer(df_Train,model,vocab1,vocab2)
#Get predictions
result = modelmaker.model_tester(df_Test,trained_model,trained_vocab,tagged_vocab)
return trained_model, trained_vocab, tagged_vocab, result, df_Test
def Train_A_Model_Direct(tag, model, vocab1, vocab2, df_Train, df_Test):
#Train the actual model:
trained_model, trained_vocab, tagged_vocab = modelmaker.model_trainer(
df_Train, model, vocab1, vocab2)
#Get predictions
result = modelmaker.model_tester(df_Test, trained_model, trained_vocab,
tagged_vocab)
return trained_model, trained_vocab, tagged_vocab, result, df_Test
def Train_A_Model(tag, model, vocab1,vocab2,myconfig,downsample=True):
df_all = dfmaker.get_train_dfs(tag,myconfig)
df_Train, df_Test = dfmaker.GenerateTestTrain(df_all)
#Train the actual model:
trained_model,trained_vocab,tagged_vocab = modelmaker.model_trainer(df_Train,model,vocab1,vocab2,downsample)
#Get predictions
result = modelmaker.model_tester(df_Test,trained_model,trained_vocab,tagged_vocab)
return trained_model, trained_vocab, tagged_vocab, result, df_Test
def Train_A_Model(tag, model, vocab1, vocab2, myconfig, downsample=True):
df_all = dfmaker.get_train_dfs(tag, myconfig)
df_Train, df_Test = dfmaker.GenerateTestTrain(df_all)
#Train the actual model:
trained_model, trained_vocab, tagged_vocab = modelmaker.model_trainer(
df_Train, model, vocab1, vocab2, downsample)
#Get predictions
result = modelmaker.model_tester(df_Test, trained_model, trained_vocab,
tagged_vocab)
return trained_model, trained_vocab, tagged_vocab, result, df_Test
def Train_A_Model_Direct(tag, model, vocab1,vocab2,df_Train,df_Test):
"""Given a testing and training dataframe, train a model
tag = tag name
model = model object (sklearn) to train to
vocab1 = Body vocab
vocab2 = Tag vocab
df_Train = Train dataframe
df_Test = Test dataframe
"""
#Train the actual model:
trained_model,trained_vocab,tagged_vocab = modelmaker.model_trainer(df_Train,model,vocab1,vocab2)
#Get predictions
result = modelmaker.model_tester(df_Test,trained_model,trained_vocab,tagged_vocab)
#Return trained model, vocab, prediction:
return trained_model, trained_vocab, tagged_vocab, result, df_Test
def Train_A_Model(tag, model, vocab1,vocab2,myconfig,downsample=True):
"""Given a tag name, train a model
tag = tag name
model = model object (sklearn) to train to
vocab1 = Body vocab
vocab2 = Tag vocab
myconfig = config file for accessing MySQL database
downsample = force spoiler and non-spoiler sets to have same size
"""
#Get the data and make Test/Train frames:
df_all = dfmaker.get_train_dfs(tag,myconfig)
df_Train, df_Test = dfmaker.GenerateTestTrain(df_all)
#Train the actual model:
trained_model,trained_vocab,tagged_vocab = modelmaker.model_trainer(df_Train,model,vocab1,vocab2,downsample)
#Get predictions
result = modelmaker.model_tester(df_Test,trained_model,trained_vocab,tagged_vocab)
#Return trained model, vocab, prediction:
return trained_model, trained_vocab, tagged_vocab, result, df_Test