def precomp_prediction(data_train, data_test, outfile, lamb=0.3):
lin_l = LinearLearner(lamb)
lin_c = lin_l(data_train)
out_file = open(outfile, "wt")
for row in data_test:
out_file.write(lpputils.tsadd(row[DEP_IDX], lin_c(row)) + "\n")
out_file.close()
def comp_prediction(data_train, data_test, outfile, lamb=0.3):
lin_l = LineSpecificLearner(LinearLearner(lamb))
lin_c = lin_l(data_train)
#rf_l = LineSpecificLearner(RFRegressorLearner())
#rf_c = rf_l(data_train)
out_file = open(outfile, "wt")
for row in data_test:
out_file.write(lpputils.tsadd(row[DEP_IDX], lin_c(row)) + "\n")
out_file.close()
def comp_prediction_neural(data_train, data_test, outfile):
X_train, y_train = prepare_NN_data(data_train, False)
X_test, _ = prepare_NN_data(data_test, True)
mlpr = MLPRegressor(hidden_layer_sizes=(300,),activation='tanh')
mlpr.fit(X_train,y_train)
pred = mlpr.predict(X_test)
print("Mreza narejena ....")
out_file = open(outfile, "wt")
st = 0
for row in data_test:
out_file.write(lpputils.tsadd(row[DEP_IDX],pred[st]) + "\n")
st += 1
out_file.close()
else:
linije[primer[3]] = [primer]
linearna_regresija = linear.LinearLearner()
for linija in linije.keys():
x, y = zgradi_matrike(linije[linija], True)
linije[linija] = linearna_regresija(x, y)
f = gzip.open("test.csv.gz", "rt",
encoding="latin1") #za izpis MAE spremeni v "train.csv.gz"
vrstica = csv.reader(f, delimiter="\t")
next(vrstica)
ime, dejanski_cas, primeri, testni_X = zgradi_matrike(vrstica, False)
datoteka = open("napovedi_tekmovanje.txt", "wt", encoding="latin1")
mae_mesec = 11
mae = 0
stevilo_primerov = 0
for vrstica in range(len(primeri)):
napoved = lpputils.tsadd(primeri[vrstica],
-linije[ime[vrstica]](testni_X[vrstica]))
datoteka.write(napoved + "\n")
if lpputils.parsedate(primeri[vrstica]).month == mae_mesec:
mae += absolute_error(dejanski_cas[vrstica], napoved)
stevilo_primerov += 1
datoteka.close()
if mae != 0:
print("Mean absolute error:", mae / stevilo_primerov)
class AverageTripClassifier(object):
def __init__(self, mean):
self.mean = mean
def __call__(self, x):
# does not use the input example at all, because
# in this case the prediction is always the same
return self.mean
if __name__ == "__main__":
f = gzip.open("train_pred.csv.gz", "rt")
reader = csv.reader(f, delimiter="\t")
next(reader)
data = [d for d in reader]
l = AverageTripLearner()
c = l(data)
f = gzip.open("test_pred.csv.gz", "rt")
reader = csv.reader(f, delimiter="\t")
next(reader) # skip legend
fo = open("naloga3.txt", "wt")
for l in reader:
print(l[-3])
fo.write(lpputils.tsadd(l[-3], c(l)) + "\n")
fo.close()
import gzip
import csv
import lpputils
f = gzip.open("test_pred.csv.gz", "rt")
reader = csv.reader(f, delimiter="\t")
next(reader) #skip legend
fo = open("polurniki.txt", "wt")
for l in reader:
fo.write(lpputils.tsadd(l[-3], 30 * 60) + "\n")
fo.close()
dejanski_cas = []
for d in reader:
originalen_datum.append(d[-3])
dejanski_cas.append(d[-1])
x.append(napolni_x(d, prazniki))
X = linear.append_ones(np.array(x))
return dejanski_cas, originalen_datum, X
X, Y = zgradi_matrike("train_pred.csv.gz", True)
linearna_regresija = linear.LinearLearner()
predikcijski_model = linearna_regresija(X, Y)
dejanski_cas, primeri, testni_X = zgradi_matrike(
"test_pred.csv.gz", False) #za izpis MAE spremeni v "train_pred.csv.gz"
mae_mesec = 11 #MAE za november
mae = 0
stevilo_primerov = 0
datoteka = open("napovedi.txt", "wt", encoding="latin1")
for vrstica in range(len(primeri)):
napoved = lpputils.tsadd(primeri[vrstica],
-predikcijski_model(testni_X[vrstica]))
datoteka.write(napoved + "\n")
if lpputils.parsedate(primeri[vrstica]).month == mae_mesec:
mae += absolute_error(dejanski_cas[vrstica], napoved)
stevilo_primerov += 1
datoteka.close()
if mae != 0:
print("Mean absolute error:", mae / stevilo_primerov)