Mr-Geekman

from typing import List
from typing import Optional

import numpy as np
import pandas as pd
from etna.models.base import BaseAdapter
from etna.models.base import MultiSegmentModel
from sklearn.base import ClassifierMixin

Mr-Geekman

target,segment,timestamp
1149.87,m1,2012-03-01
1053.8002,m1,2012-04-01
1388.8798,m1,2012-05-01
1783.3702,m1,2012-06-01
1921.0252,m1,2012-07-01
2704.9449,m1,2012-08-01
4184.4135,m1,2012-09-01
4148.3542,m1,2012-10-01
2620.7251,m1,2012-11-01

Mr-Geekman

import pandas as pd
from etna.datasets import TSDataset


df = pd.read_csv(DATA_PATH / "tourism_monthly.csv", parse_dates=["timestamp"])
ts = TSDataset(df=TSDataset.to_dataset(df), freq="MS")

Mr-Geekman

>>> ts.info()
<class 'etna.datasets.TSDataset'>
num_segments: 366
num_exogs: 0
num_regressors: 0
num_known_future: 0
freq: MS

>>> len(ts.index)
333

Mr-Geekman

import time

from etna.metrics import SMAPE
from etna.models import CatBoostMultiSegmentModel
from etna.pipeline import Pipeline
from etna.transforms import LagTransform
from etna.transforms import SegmentEncoderTransform

HORIZON = 24
N_FOLDS = 3

Mr-Geekman

>>> baseline_result = get_baseline_metrics(ts=ts, horizon=HORIZON, n_folds=N_FOLDS)
>>> print(f"SMAPE: {baseline_result['SMAPE']:.2f}")
SMAPE: 27.03
>>> print(f"Time: {baseline_result['time']:.2f}s")
Time: 46.16s

Mr-Geekman

def get_metrics_by_horizon(ts: TSDataset, horizon: int, total_points: int):
    if total_points % horizon != 0:
        raise ValueError("Horizon should divide total number of points!")

    n_folds = total_points // horizon

    transforms = [
        LagTransform(
            in_column="target",
            lags=list(range(horizon, MAX_LAG)),

Mr-Geekman

from etna.pipeline import AutoRegressivePipeline


def get_metrics_by_step(ts: TSDataset, step: int, horizon: int, n_folds: int):
    transforms = [
        LagTransform(
            in_column="target",
            lags=list(range(step, MAX_LAG)),
            out_column="lag_target",
        ),

Mr-Geekman

from etna.ensembles import DirectEnsemble
from etna.pipeline import assemble_pipelines


def get_metrics_by_horizon_step(
    ts: TSDataset, horizon_step: int, horizon: int, n_folds: int
):
    if horizon % horizon_step != 0:
        raise ValueError("Horizon step should devide horizon!")