Target encoding
Memenangi Kompetisi Kaggle dengan Python
Yauhen Babakhin
Kaggle Grandmaster
Fitur kategorikal berkardinalitas tinggi
- Label encoder memberi nomor unik untuk tiap kategori
- One-hot encoder membuat fitur baru untuk tiap nilai kategori
- Target encoding jadi solusinya!
Mean target encoding
| ID Train | Kategorikal | Target |
|---|---|---|
| 1 | A | 1 |
| 2 | B | 0 |
| 3 | B | 0 |
| 4 | A | 1 |
| 5 | B | 0 |
| 6 | A | 0 |
| 7 | B | 1 |
| ID Test | Kategorikal | Target |
|---|---|---|
| 10 | A | ? |
| 11 | A | ? |
| 12 | B | ? |
| 13 | A | ? |
Mean target encoding
- Hitung rata-rata pada train, terapkan ke test
- Bagi train jadi K fold. Hitung rata-rata pada (K-1) fold, terapkan ke fold ke-K
- Tambahkan fitur mean target encoded ke model
Hitung rata-rata pada train
| ID Train | Kategorikal | Target |
|---|---|---|
| 1 | A | 1 |
| 2 | B | 0 |
| 3 | B | 0 |
| 4 | A | 1 |
| 5 | B | 0 |
| 6 | A | 0 |
| 7 | B | 1 |
Hitung rata-rata pada train
| ID Train | Kategorikal | Target |
|---|---|---|
| 1 | A | 1 |
| 2 | B | 0 |
| 3 | B | 0 |
| 4 | A | 1 |
| 5 | B | 0 |
| 6 | A | 0 |
| 7 | B | 1 |
Hitung rata-rata pada train
| ID Train | Kategorikal | Target |
|---|---|---|
| 1 | A | 1 |
| 2 | B | 0 |
| 3 | B | 0 |
| 4 | A | 1 |
| 5 | B | 0 |
| 6 | A | 0 |
| 7 | B | 1 |
Encoding pada test
| ID Test | Kategorikal | Target | Mean encoded |
|---|---|---|---|
| 10 | A | ? | 0.66 |
| 11 | A | ? | 0.66 |
| 12 | B | ? | 0.25 |
| 13 | A | ? | 0.66 |
Encoding train dengan out-of-fold
| ID Train | Kategorikal | Target | Fold |
|---|---|---|---|
| 1 | A | 1 | 1 |
| 2 | B | 0 | 1 |
| 3 | B | 0 | 1 |
| 4 | A | 1 | 1 |
| 5 | B | 0 | 2 |
| 6 | A | 0 | 2 |
| 7 | B | 1 | 2 |
Encoding train dengan out-of-fold
| ID Train | Kategorikal | Target | Fold | Mean encoded |
|---|---|---|---|---|
| 1 | A | 1 | 1 | |
| 2 | B | 0 | 1 | |
| 3 | B | 0 | 1 | |
| 4 | A | 1 | 1 | |
| 5 | B | 0 | 2 | |
| 6 | A | 0 | 2 | |
| 7 | B | 1 | 2 |
Encoding train dengan out-of-fold
| ID Train | Kategorikal | Target | Fold | Mean encoded |
|---|---|---|---|---|
| 1 | A | 1 | 1 | 0 |
| 2 | B | 0 | 1 | 0.5 |
| 3 | B | 0 | 1 | 0.5 |
| 4 | A | 1 | 1 | 0 |
| 5 | B | 0 | 2 | |
| 6 | A | 0 | 2 | |
| 7 | B | 1 | 2 |
Encoding train dengan out-of-fold
| ID Train | Kategorikal | Target | Fold | Mean encoded |
|---|---|---|---|---|
| 1 | A | 1 | 1 | 0 |
| 2 | B | 0 | 1 | 0.5 |
| 3 | B | 0 | 1 | 0.5 |
| 4 | A | 1 | 1 | 0 |
| 5 | B | 0 | 2 | |
| 6 | A | 0 | 2 | |
| 7 | B | 1 | 2 |
Encoding train dengan out-of-fold
| ID Train | Kategorikal | Target | Fold | Mean encoded |
|---|---|---|---|---|
| 1 | A | 1 | 1 | 0 |
| 2 | B | 0 | 1 | 0.5 |
| 3 | B | 0 | 1 | 0.5 |
| 4 | A | 1 | 1 | 0 |
| 5 | B | 0 | 2 | 0 |
| 6 | A | 0 | 2 | 1 |
| 7 | B | 1 | 2 | 0 |
Panduan praktis
Panduan praktis
Smoothing
$$mean\_enc_i = \frac{target\_sum_i}{n_i}$$
$$smoothed\_mean\_enc_i = \frac{target\_sum_i + \alpha*global\_mean}{n_i + \alpha}$$
$$\alpha \in [5; 10]$$
Panduan praktis
Smoothing
$$mean\_enc_i = \frac{target\_sum_i}{n_i}$$
$$smoothed\_mean\_enc_i = \frac{target\_sum_i + \alpha*global\_mean}{n_i + \alpha}$$
$$\alpha \in [5; 10]$$
Kategori baru
- Isi kategori baru di data test dengan global_mean
Panduan praktis
| ID Train | Kategorikal | Target |
|---|---|---|
| 1 | A | 1 |
| 2 | B | 0 |
| 3 | B | 0 |
| 4 | A | 0 |
| 5 | B | 1 |
| ID Test | Kategorikal | Target | Mean encoded |
|---|---|---|---|
| 10 | A | ? | 0.43 |
| 11 | B | ? | 0.38 |
| 12 | C | ? | 0.40 |
Ayo berlatih!
Memenangi Kompetisi Kaggle dengan Python
