Penskalaan data dan Regresi KNN

Machine Learning untuk Keuangan dengan Python

Nathan George

Data Science Professor

kepentingan fitur

Seleksi fitur: hapus hari kerja

print(feature_names)

['10d_close_pct',
 '14-day SMA',
 '14-day RSI',
 '200-day SMA',
 '200-day RSI',
 'Adj_Volume_1d_change',
 'Adj_Volume_1d_change_SMA',
 'weekday_1',
 'weekday_2',
 'weekday_3',
 'weekday_4']

print(feature_names[:-4])

['10d_close_pct',
 '14-day SMA',
 '14-day RSI',
 '200-day SMA',
 '200-day RSI',
 'Adj_Volume_1d_change',
 'Adj_Volume_1d_change_SMA']

Hapus hari kerja

train_features = train_features.iloc[:, :-4]
test_features = test_features.iloc[:, :-4]

plot fitur 2D

prediksi knn tidak dikenal

prediksi knn dengan 2 titik terdekat

jarak minowski

fitur besar dan kecil

Opsi penskalaan

Opsi penskalaan:

min-max
standardisasi
median-MAD
pemetaan ke fungsi sewenang-wenang (mis. sigmoid, tanh)

plot fitur 2D sebelum dan sesudah penskalaan

scale dari sklearn

from sklearn.preprocessing import scale

sc = scale()
scaled_train_features = sc.fit_transform(train_features)
scaled_test_features = sc.transform(test_features)

sebelum dan sesudah standardisasi

Membuat subplot

# create figure and list containing axes
f, ax = plt.subplots(nrows=2, ncols=1)

# plot histograms of before and after scaling
train_features.iloc[:, 2].hist(ax=ax[0])
ax[1].hist(scaled_train_features[:, 2])
plt.show()

Skalakan data dan gunakan KNN!

Machine Learning untuk Keuangan dengan Python