def fetchNewMatches():
db.create_database(dbfile) # create db if not exists
classificationmodel = ml.load_model(
classificationmodelname) # load model for classification
regressionmodel = ml.load_model(
regressionmodelname) # load model for regression
for match in fetching.fetchseason(
season): # for each past match of season 2019-2020
hometeamdata = db.get_teamhistory(dbfile, match["home-team"],
match["date"]) # get history
awayteamdata = db.get_teamhistory(dbfile, match["away-team"],
match["date"]) # get history
prepared_data = ml.prepare_matchdata(
match, hometeamdata, awayteamdata) # prepare data for model
predictedClassificationResult = ml.exec_classification_model(
classificationmodel, prepared_data) # run the model
predictedRegressionResult = ml.exec_regression_model(
regressionmodel, prepared_data) # run the model
db.add_match(dbfile, match["home-team"], match["away-team"],
match["date"], predictedClassificationResult,
predictedRegressionResult, match["odds-home"],
match["odds-draw"], match["odds-away"],
match["home-goals"], match["home-shots"],
match["home-shots-on-target"], match["away-goals"],
match["away-shots"], match["away-shots-on-target"])
return "ok"
def retrainRegression():
matches = []
for match in db.get_matches(dbfile): # for each match in db
hometeamdata = db.get_teamhistory(dbfile, match["home-team"],
match["date"]) # get history
awayteamdata = db.get_teamhistory(dbfile, match["away-team"],
match["date"]) # get history
prepared_data = ml.prepare_regressiontrainingmatchdata(
match, hometeamdata, awayteamdata) # prepare data for model
matches.append(prepared_data)
model = ml.retrain_regression_model(matches) # build&create model
ml.save_model(model, regressionmodelname) # save model
# repredict everything!
for match in db.get_matches(dbfile):
hometeamdata = db.get_teamhistory(dbfile, match["home-team"],
match["date"]) # get history
awayteamdata = db.get_teamhistory(dbfile, match["away-team"],
match["date"]) # get history
prepared_data = ml.prepare_matchdata(
match, hometeamdata, awayteamdata) # prepare data for model
predictedRegressionResult = ml.exec_regression_model(
model, prepared_data) # run the model
db.update_matchprediction_regression(dbfile, match["home-team"],
match["away-team"], match["date"],
predictedRegressionResult)
return "ok"
def predict_match(classificationmodel, regressionmodel, match):
hometeamdata = db.get_teamhistory(dbfile, match[4],
match[2]) # get history
awayteamdata = db.get_teamhistory(dbfile, match[5],
match[2]) # get history
prepared_data = ml.prepare_matchdata(
match[2], hometeamdata, awayteamdata) # prepare data for model
predictedClassificationResult = ml.exec_classification_model(
classificationmodel, prepared_data) # run the model
predictedRegressionResult = ml.exec_regression_model(
regressionmodel, prepared_data) # run the model
return predictedClassificationResult, predictedRegressionResult
def main():
db.create_database(dbfile)
fetchNewMatches()
dbmatch = db.get_match(dbfile, "sc-paderborn-07", "borussia-dortmund",
20200601)
hometeamdata = db.get_teamhistory(dbfile, "sc-paderborn-07", 20200601)
awayteamdata = db.get_teamhistory(dbfile, "borussia-dortmund", 20200601)
jasonstring = ml.prepare_matchdata(dbmatch, hometeamdata,
awayteamdata) # prepare data for model
df = pd.read_json(jasonstring).T # read json and create dataframe
model = ml.load_model("model02_H3_M") # load model
predictedResultArray = ml.exec_classification_model(
model, df) # retrieve result from model
predictedResult = numpy.argmax(predictedResultArray,
axis=None) # get max value
print(predictedResult)