mltree 0.0.2

A machine learning package only for tree based models

mltree

Install

pip install mltree

How to use

First, load the analytical base table:

from mltree.train import train_tree_models

import pandas as pd
from pathlib import Path

path = Path('..')
datasets_path = path/'datasets'

df = pd.read_csv(datasets_path/'churn_abt.csv')

df.head()

<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; }

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	data_ref_safra	seller_id	uf	tot_orders_12m	tot_items_12m	tot_items_dist_12m	receita_12m	recencia	nao_revendeu_next_6m
0	2018-01-01	0015a82c2db000af6aaaf3ae2ecb0532	SP	3	3	1	2685.00	74	1
1	2018-01-01	001cca7ae9ae17fb1caed9dfb1094831	ES	171	207	9	21275.23	2	0
2	2018-01-01	002100f778ceb8431b7a1020ff7ab48f	SP	38	42	15	781.80	2	0
3	2018-01-01	003554e2dce176b5555353e4f3555ac8	GO	1	1	1	120.00	16	1
4	2018-01-01	004c9cd9d87a3c30c522c48c4fc07416	SP	130	141	75	16228.88	8	0

Split into train and test or out of time datasets:

df_train = df.query('data_ref_safra < "2018-03-01"')
df_oot = df.query('data_ref_safra == "2018-03-01"')

Get features metadata and types:

key_vars = ['data_ref_safra', 'seller_id']
target = 'nao_revendeu_next_6m'
num_vars = [ var for var in df.select_dtypes(include='number').columns.tolist() if var not in [target] ]
cat_vars = [var for var in df.select_dtypes(exclude='number').columns.tolist() if var not in key_vars]

Train based tree models:

train_tree_models(df_train, df_oot, target=target, folds=5, cat_features=cat_vars, num_features=num_vars, seed=42)

{'dt': {'auc': {'train': 0.9139680595991275, 'test': 0.8968114296299949}},
 'rf': {'auc': {'train': 0.9072972070544887, 'test': 0.8964968670043654}}}