![](\"https://mybinder.org/badge_logo.svg\"){kind=icon}
\n",
"\n",
""
]
},
{
"cell_type": "markdown",
"id": "bca01a6c",
"metadata": {},
"source": [
"This notebooks covers AutoML utilities of ETNA library.\n",
"\n",
"- [Hyperparameters tuning](#chapter_1)\n",
" - [How `Tune` works](#section_1_1)\n",
" - [Example](#section_1_2)\n",
"- [General AutoML](#chapter_2)\n",
" - [How `Auto` works](#section_2_1)\n",
" - [Example](#section_2_2)"
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "6f70e872",
"metadata": {},
"outputs": [],
"source": [
"import warnings\n",
"\n",
"\n",
"warnings.filterwarnings(\"ignore\")"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "e50060f6",
"metadata": {},
"outputs": [],
"source": [
"HORIZON = 14"
]
},
{
"cell_type": "markdown",
"id": "e5c2c568",
"metadata": {},
"source": [
"Many features from this notebook require `auto` extension. You can install it by the command:\n",
"```bash\n",
"pip install etna[auto]\n",
"```"
]
},
{
"cell_type": "markdown",
"id": "33ad7417",
"metadata": {},
"source": [
"## 1. Hyperparameters tuning "
]
},
{
"cell_type": "markdown",
"id": "4542c8eb",
"metadata": {},
"source": [
"It is a common task to tune hyperparameters of existing pipeline to improve its quality. For this purpose there is an `etna.auto.Tune` class, which is responsible for creating [optuna](https://github.com/optuna/optuna) study to solve this problem.\n",
"\n",
"In the next sections we will see how it works and how to use it for your particular problems."
]
},
{
"cell_type": "markdown",
"id": "73194640",
"metadata": {},
"source": [
"### 1.1 How `Tune` works "
]
},
{
"cell_type": "markdown",
"id": "c7777ea3",
"metadata": {},
"source": [
"During init `Tune` accepts `pipeline`, its tuning parameters (`params_to_tune`), optimization metric (`target_metric`), parameters of backtest and parameters of optuna study.\n",
"\n",
"In `fit` the optuna study is created. During each trial the sample of parameters is generated from `params_to_tune` and applied to `pipeline`. After that, the new pipeline is checked in backtest and target metric is returned to optuna framework."
]
},
{
"cell_type": "markdown",
"id": "09e6cb8e",
"metadata": {},
"source": [
"Let's look closer at `params_to_tune` parameter. It expects dictionary with parameter names and its distributions. But how this parameter names should be chosen?"
]
},
{
"cell_type": "markdown",
"id": "5d7a777a",
"metadata": {},
"source": [
"#### 1.1.1 `set_params`"
]
},
{
"cell_type": "markdown",
"id": "cc05b85b",
"metadata": {},
"source": [
"We are going to make a little detour to explain the `set_params` method, which is supported by ETNA pipelines, models and transforms. Given a dictionary with parameters it allows to create from existing object a new one with changed parameters."
]
},
{
"cell_type": "markdown",
"id": "b291efa4",
"metadata": {},
"source": [
"First, we define some objects for our future examples."
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "9d6893b8",
"metadata": {},
"outputs": [],
"source": [
"from etna.pipeline import Pipeline\n",
"from etna.models import LinearPerSegmentModel\n",
"from etna.transforms import LagTransform\n",
"from etna.transforms import DateFlagsTransform\n",
"\n",
"\n",
"model = LinearPerSegmentModel()\n",
"transforms = [\n",
" LagTransform(in_column=\"target\", lags=list(range(HORIZON, HORIZON + 10)), out_column=\"target_lag\"),\n",
" DateFlagsTransform(out_column=\"date_flags\"),\n",
"]\n",
"pipeline = Pipeline(model=model, transforms=transforms, horizon=HORIZON)"
]
},
{
"cell_type": "markdown",
"id": "01a57e5c",
"metadata": {},
"source": [
"Let's look at simple example, when we want to change `fit_intercept` parameter of the `model`."
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "32c51370",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"{'fit_intercept': True,\n",
" 'kwargs': {},\n",
" '_target_': 'etna.models.linear.LinearPerSegmentModel'}"
]
},
"execution_count": 4,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"model.to_dict()"
]
},
{
"cell_type": "code",
"execution_count": 5,
"id": "60bc963f",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"{'fit_intercept': False,\n",
" 'kwargs': {},\n",
" '_target_': 'etna.models.linear.LinearPerSegmentModel'}"
]
},
"execution_count": 5,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"new_model_params = {\"fit_intercept\": False}\n",
"new_model = model.set_params(**new_model_params)\n",
"new_model.to_dict()"
]
},
{
"cell_type": "markdown",
"id": "383931c2",
"metadata": {},
"source": [
"Great! On the next step we want to change the `fit_intercept` of `model` inside the `pipeline`."
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "7ff49f9a",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"{'model': {'fit_intercept': True,\n",
" 'kwargs': {},\n",
" '_target_': 'etna.models.linear.LinearPerSegmentModel'},\n",
" 'transforms': [{'in_column': 'target',\n",
" 'lags': [14, 15, 16, 17, 18, 19, 20, 21, 22, 23],\n",
" 'out_column': 'target_lag',\n",
" '_target_': 'etna.transforms.math.lags.LagTransform'},\n",
" {'day_number_in_week': True,\n",
" 'day_number_in_month': True,\n",
" 'day_number_in_year': False,\n",
" 'week_number_in_month': False,\n",
" 'week_number_in_year': False,\n",
" 'month_number_in_year': False,\n",
" 'season_number': False,\n",
" 'year_number': False,\n",
" 'is_weekend': True,\n",
" 'special_days_in_week': (),\n",
" 'special_days_in_month': (),\n",
" 'out_column': 'date_flags',\n",
" '_target_': 'etna.transforms.timestamp.date_flags.DateFlagsTransform'}],\n",
" 'horizon': 14,\n",
" '_target_': 'etna.pipeline.pipeline.Pipeline'}"
]
},
"execution_count": 6,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"pipeline.to_dict()"
]
},
{
"cell_type": "code",
"execution_count": 7,
"id": "497662b6",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"{'model': {'fit_intercept': False,\n",
" 'kwargs': {},\n",
" '_target_': 'etna.models.linear.LinearPerSegmentModel'},\n",
" 'transforms': [{'in_column': 'target',\n",
" 'lags': [14, 15, 16, 17, 18, 19, 20, 21, 22, 23],\n",
" 'out_column': 'target_lag',\n",
" '_target_': 'etna.transforms.math.lags.LagTransform'},\n",
" {'day_number_in_week': True,\n",
" 'day_number_in_month': True,\n",
" 'day_number_in_year': False,\n",
" 'week_number_in_month': False,\n",
" 'week_number_in_year': False,\n",
" 'month_number_in_year': False,\n",
" 'season_number': False,\n",
" 'year_number': False,\n",
" 'is_weekend': True,\n",
" 'special_days_in_week': (),\n",
" 'special_days_in_month': (),\n",
" 'out_column': 'date_flags',\n",
" '_target_': 'etna.transforms.timestamp.date_flags.DateFlagsTransform'}],\n",
" 'horizon': 14,\n",
" '_target_': 'etna.pipeline.pipeline.Pipeline'}"
]
},
"execution_count": 7,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"new_pipeline_params = {\"model.fit_intercept\": False}\n",
"new_pipeline = pipeline.set_params(**new_pipeline_params)\n",
"new_pipeline.to_dict()"
]
},
{
"cell_type": "markdown",
"id": "8eba262b",
"metadata": {},
"source": [
"Ok, it looks like we managed to do this. On the last step we are going to change `is_weekend` flag of `DateFlagsTransform` inside our `pipeline`."
]
},
{
"cell_type": "code",
"execution_count": 8,
"id": "28a1ac00",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"{'model': {'fit_intercept': True,\n",
" 'kwargs': {},\n",
" '_target_': 'etna.models.linear.LinearPerSegmentModel'},\n",
" 'transforms': [{'in_column': 'target',\n",
" 'lags': [14, 15, 16, 17, 18, 19, 20, 21, 22, 23],\n",
" 'out_column': 'target_lag',\n",
" '_target_': 'etna.transforms.math.lags.LagTransform'},\n",
" {'day_number_in_week': True,\n",
" 'day_number_in_month': True,\n",
" 'day_number_in_year': False,\n",
" 'week_number_in_month': False,\n",
" 'week_number_in_year': False,\n",
" 'month_number_in_year': False,\n",
" 'season_number': False,\n",
" 'year_number': False,\n",
" 'is_weekend': False,\n",
" 'special_days_in_week': (),\n",
" 'special_days_in_month': (),\n",
" 'out_column': 'date_flags',\n",
" '_target_': 'etna.transforms.timestamp.date_flags.DateFlagsTransform'}],\n",
" 'horizon': 14,\n",
" '_target_': 'etna.pipeline.pipeline.Pipeline'}"
]
},
"execution_count": 8,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"new_pipeline_params = {\"transforms.1.is_weekend\": False}\n",
"new_pipeline = pipeline.set_params(**new_pipeline_params)\n",
"new_pipeline.to_dict()"
]
},
{
"cell_type": "markdown",
"id": "50c2bf0d",
"metadata": {},
"source": [
"As we can see, we managed to do this."
]
},
{
"cell_type": "markdown",
"id": "4deb8b8b",
"metadata": {},
"source": [
"#### 1.1.2 `params_to_tune`"
]
},
{
"cell_type": "markdown",
"id": "8c732645",
"metadata": {},
"source": [
"Let's get back to our initial question about `params_to_tune`. In our optuna study we are going to sample each parameter value from its distribution and pass it into `pipeline.set_params` method. So, the keys for `params_to_tune` should be a valid for `set_params` method.\n",
"\n",
"Distributions are taken from `etna.distributions` and they are matching `optuna.Trial.suggest_` methods."
]
},
{
"cell_type": "markdown",
"id": "f6a39f16",
"metadata": {},
"source": [
"For example, something like this will be valid for our `pipeline` defined above:"
]
},
{
"cell_type": "code",
"execution_count": 9,
"id": "4dab566f",
"metadata": {},
"outputs": [],
"source": [
"from etna.distributions import CategoricalDistribution\n",
"\n",
"\n",
"example_params_to_tune = {\n",
" \"model.fit_intercept\": CategoricalDistribution([False, True]),\n",
" \"transforms.0.is_weekend\": CategoricalDistribution([False, True]),\n",
"}"
]
},
{
"cell_type": "markdown",
"id": "a0c81b84",
"metadata": {},
"source": [
"There are some good news: it isn't necessary for our users to define `params_to_tune`, because we have a default grid for many of our classes. The default grid is available by calling `params_to_tune` method on pipeline, model or transform. Let's check our `pipeline`:"
]
},
{
"cell_type": "code",
"execution_count": 10,
"id": "b493dace",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"{'model.fit_intercept': CategoricalDistribution(choices=[False, True]),\n",
" 'transforms.1.day_number_in_week': CategoricalDistribution(choices=[False, True]),\n",
" 'transforms.1.day_number_in_month': CategoricalDistribution(choices=[False, True]),\n",
" 'transforms.1.day_number_in_year': CategoricalDistribution(choices=[False, True]),\n",
" 'transforms.1.week_number_in_month': CategoricalDistribution(choices=[False, True]),\n",
" 'transforms.1.week_number_in_year': CategoricalDistribution(choices=[False, True]),\n",
" 'transforms.1.month_number_in_year': CategoricalDistribution(choices=[False, True]),\n",
" 'transforms.1.season_number': CategoricalDistribution(choices=[False, True]),\n",
" 'transforms.1.year_number': CategoricalDistribution(choices=[False, True]),\n",
" 'transforms.1.is_weekend': CategoricalDistribution(choices=[False, True])}"
]
},
"execution_count": 10,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"pipeline.params_to_tune()"
]
},
{
"cell_type": "markdown",
"id": "554c5af2",
"metadata": {},
"source": [
"Now we are ready to use it in practice."
]
},
{
"cell_type": "markdown",
"id": "df2102f8",
"metadata": {},
"source": [
"### 1.2 Example "
]
},
{
"cell_type": "markdown",
"id": "535c0b18",
"metadata": {},
"source": [
"#### 1.2.1 Loading data"
]
},
{
"cell_type": "markdown",
"id": "9352eeb4",
"metadata": {},
"source": [
"Let's start by loading example data."
]
},
{
"cell_type": "code",
"execution_count": 11,
"id": "1b4e7705",
"metadata": {},
"outputs": [],
"source": [
"import pandas as pd\n",
"\n",
"from etna.datasets import TSDataset"
]
},
{
"cell_type": "code",
"execution_count": 12,
"id": "0041c9ab",
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"| \n", " | timestamp | \n", "segment | \n", "target | \n", "
|---|---|---|---|
| 0 | \n", "2019-01-01 | \n", "segment_a | \n", "170 | \n", "
| 1 | \n", "2019-01-02 | \n", "segment_a | \n", "243 | \n", "
| 2 | \n", "2019-01-03 | \n", "segment_a | \n", "267 | \n", "
| 3 | \n", "2019-01-04 | \n", "segment_a | \n", "287 | \n", "
| 4 | \n", "2019-01-05 | \n", "segment_a | \n", "279 | \n", "
| \n", " | pipeline | \n", "hash | \n", "Sign_median | \n", "Sign_mean | \n", "Sign_std | \n", "Sign_percentile_5 | \n", "Sign_percentile_25 | \n", "Sign_percentile_75 | \n", "Sign_percentile_95 | \n", "SMAPE_median | \n", "... | \n", "MSE_percentile_75 | \n", "MSE_percentile_95 | \n", "MedAE_median | \n", "MedAE_mean | \n", "MedAE_std | \n", "MedAE_percentile_5 | \n", "MedAE_percentile_25 | \n", "MedAE_percentile_75 | \n", "MedAE_percentile_95 | \n", "state | \n", "
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "f4f02e1d5f60b8f322a4a8a622dd1c1e | \n", "-0.500000 | \n", "-0.478571 | \n", "0.205204 | \n", "-0.672857 | \n", "-0.621429 | \n", "-0.357143 | \n", "-0.254286 | \n", "5.806429 | \n", "... | \n", "2220.282484 | \n", "2953.865443 | \n", "21.000232 | \n", "22.334611 | \n", "8.070926 | \n", "14.955846 | \n", "18.861388 | \n", "24.473455 | \n", "31.581505 | \n", "TrialState.COMPLETE | \n", "
| 1 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "3d7b7af16d71a36f3b935f69e113e22d | \n", "-0.457143 | \n", "-0.485714 | \n", "0.242437 | \n", "-0.745714 | \n", "-0.642857 | \n", "-0.300000 | \n", "-0.265714 | \n", "5.856039 | \n", "... | \n", "2644.982216 | \n", "3294.855806 | \n", "22.762122 | \n", "23.389796 | \n", "8.482028 | \n", "14.897792 | \n", "19.344439 | \n", "26.807479 | \n", "32.760543 | \n", "TrialState.COMPLETE | \n", "
| 2 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "7c7932114268832a5458acfecfb453fc | \n", "-0.200000 | \n", "-0.271429 | \n", "0.264447 | \n", "-0.581429 | \n", "-0.392857 | \n", "-0.078571 | \n", "-0.061429 | \n", "5.693983 | \n", "... | \n", "3457.757162 | \n", "4209.624737 | \n", "22.572681 | \n", "23.336111 | \n", "12.049564 | \n", "11.235277 | \n", "18.503043 | \n", "27.405750 | \n", "36.505748 | \n", "TrialState.COMPLETE | \n", "
| 3 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "b7ac5f7fcf9c8959626befe263a9d561 | \n", "0.000000 | \n", "-0.085714 | \n", "0.211248 | \n", "-0.340000 | \n", "-0.100000 | \n", "0.014286 | \n", "0.048571 | \n", "7.881275 | \n", "... | \n", "5039.841145 | \n", "5665.228696 | \n", "35.976862 | \n", "33.937644 | \n", "17.252826 | \n", "14.444379 | \n", "27.282228 | \n", "42.632278 | \n", "50.576005 | \n", "TrialState.COMPLETE | \n", "
| 4 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "e928929f89156d88ef49e28abaf55847 | \n", "-0.414286 | \n", "-0.421429 | \n", "0.207840 | \n", "-0.620000 | \n", "-0.585714 | \n", "-0.250000 | \n", "-0.232857 | \n", "6.032319 | \n", "... | \n", "3091.962427 | \n", "3181.592755 | \n", "23.166650 | \n", "25.265089 | \n", "13.224461 | \n", "13.001779 | \n", "18.666844 | \n", "29.764896 | \n", "40.466215 | \n", "TrialState.COMPLETE | \n", "
| 5 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "3b4311d41fcaab7307235ea23b6d4599 | \n", "-0.400000 | \n", "-0.385714 | \n", "0.396927 | \n", "-0.788571 | \n", "-0.514286 | \n", "-0.271429 | \n", "0.037143 | \n", "6.653462 | \n", "... | \n", "3800.976318 | \n", "4837.444681 | \n", "35.792514 | \n", "32.276030 | \n", "16.296588 | \n", "13.499409 | \n", "24.106508 | \n", "43.962035 | \n", "46.129572 | \n", "TrialState.COMPLETE | \n", "
| 6 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "74065ebc11c81bed6a9819d026c7cd84 | \n", "-0.442857 | \n", "-0.435714 | \n", "0.246196 | \n", "-0.672857 | \n", "-0.621429 | \n", "-0.257143 | \n", "-0.188571 | \n", "5.739626 | \n", "... | \n", "2933.246064 | \n", "4802.299660 | \n", "27.304852 | \n", "24.936077 | \n", "8.294963 | \n", "15.108636 | \n", "21.478207 | \n", "30.762723 | \n", "31.447233 | \n", "TrialState.COMPLETE | \n", "
| 7 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "b0d0420255c6117045f8254bf8f377a0 | \n", "-0.442857 | \n", "-0.464286 | \n", "0.260167 | \n", "-0.725714 | \n", "-0.657143 | \n", "-0.250000 | \n", "-0.232857 | \n", "6.042134 | \n", "... | \n", "2682.735922 | \n", "3688.168155 | \n", "28.393903 | \n", "25.819143 | \n", "8.652993 | \n", "15.618131 | \n", "21.989342 | \n", "32.223704 | \n", "32.415490 | \n", "TrialState.COMPLETE | \n", "
| 8 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "25dcd8bb095f87a1ffc499fa6a83ef5d | \n", "-0.457143 | \n", "-0.457143 | \n", "0.265986 | \n", "-0.705714 | \n", "-0.671429 | \n", "-0.242857 | \n", "-0.208571 | \n", "5.869280 | \n", "... | \n", "3098.567787 | \n", "3154.538337 | \n", "22.380642 | \n", "24.289797 | \n", "11.998603 | \n", "13.252341 | \n", "19.168974 | \n", "27.501465 | \n", "38.000072 | \n", "TrialState.COMPLETE | \n", "
| 9 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "3f1ca1759261598081fa3bb2f32fe0ac | \n", "-0.414286 | \n", "-0.435714 | \n", "0.292654 | \n", "-0.725714 | \n", "-0.657143 | \n", "-0.192857 | \n", "-0.175714 | \n", "6.608191 | \n", "... | \n", "3044.388978 | \n", "3611.477391 | \n", "23.750327 | \n", "26.488927 | \n", "13.825791 | \n", "14.242057 | \n", "20.027917 | \n", "30.211337 | \n", "42.569838 | \n", "TrialState.COMPLETE | \n", "
| 10 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "-0.157143 | \n", "-0.185714 | \n", "0.226779 | \n", "-0.431429 | \n", "-0.328571 | \n", "-0.014286 | \n", "0.020000 | \n", "5.974832 | \n", "... | \n", "2902.306123 | \n", "3526.513999 | \n", "17.027383 | \n", "21.682156 | \n", "15.988286 | \n", "9.110958 | \n", "11.100846 | \n", "27.608693 | \n", "40.770037 | \n", "TrialState.COMPLETE | \n", "
| 11 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "-0.157143 | \n", "-0.185714 | \n", "0.226779 | \n", "-0.431429 | \n", "-0.328571 | \n", "-0.014286 | \n", "0.020000 | \n", "5.974832 | \n", "... | \n", "2902.306123 | \n", "3526.513999 | \n", "17.027383 | \n", "21.682156 | \n", "15.988286 | \n", "9.110958 | \n", "11.100846 | \n", "27.608693 | \n", "40.770037 | \n", "TrialState.COMPLETE | \n", "
| 12 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "-0.157143 | \n", "-0.185714 | \n", "0.226779 | \n", "-0.431429 | \n", "-0.328571 | \n", "-0.014286 | \n", "0.020000 | \n", "5.974832 | \n", "... | \n", "2902.306123 | \n", "3526.513999 | \n", "17.027383 | \n", "21.682156 | \n", "15.988286 | \n", "9.110958 | \n", "11.100846 | \n", "27.608693 | \n", "40.770037 | \n", "TrialState.COMPLETE | \n", "
| 13 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "-0.157143 | \n", "-0.185714 | \n", "0.226779 | \n", "-0.431429 | \n", "-0.328571 | \n", "-0.014286 | \n", "0.020000 | \n", "5.974832 | \n", "... | \n", "2902.306123 | \n", "3526.513999 | \n", "17.027383 | \n", "21.682156 | \n", "15.988286 | \n", "9.110958 | \n", "11.100846 | \n", "27.608693 | \n", "40.770037 | \n", "TrialState.COMPLETE | \n", "
| 14 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "-0.157143 | \n", "-0.185714 | \n", "0.226779 | \n", "-0.431429 | \n", "-0.328571 | \n", "-0.014286 | \n", "0.020000 | \n", "5.974832 | \n", "... | \n", "2902.306123 | \n", "3526.513999 | \n", "17.027383 | \n", "21.682156 | \n", "15.988286 | \n", "9.110958 | \n", "11.100846 | \n", "27.608693 | \n", "40.770037 | \n", "TrialState.COMPLETE | \n", "
| 15 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "-0.157143 | \n", "-0.185714 | \n", "0.226779 | \n", "-0.431429 | \n", "-0.328571 | \n", "-0.014286 | \n", "0.020000 | \n", "5.974832 | \n", "... | \n", "2902.306123 | \n", "3526.513999 | \n", "17.027383 | \n", "21.682156 | \n", "15.988286 | \n", "9.110958 | \n", "11.100846 | \n", "27.608693 | \n", "40.770037 | \n", "TrialState.COMPLETE | \n", "
| 16 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "-0.157143 | \n", "-0.185714 | \n", "0.226779 | \n", "-0.431429 | \n", "-0.328571 | \n", "-0.014286 | \n", "0.020000 | \n", "5.974832 | \n", "... | \n", "2902.306123 | \n", "3526.513999 | \n", "17.027383 | \n", "21.682156 | \n", "15.988286 | \n", "9.110958 | \n", "11.100846 | \n", "27.608693 | \n", "40.770037 | \n", "TrialState.COMPLETE | \n", "
| 17 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "-0.157143 | \n", "-0.185714 | \n", "0.226779 | \n", "-0.431429 | \n", "-0.328571 | \n", "-0.014286 | \n", "0.020000 | \n", "5.974832 | \n", "... | \n", "2902.306123 | \n", "3526.513999 | \n", "17.027383 | \n", "21.682156 | \n", "15.988286 | \n", "9.110958 | \n", "11.100846 | \n", "27.608693 | \n", "40.770037 | \n", "TrialState.COMPLETE | \n", "
| 18 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "6f595f4f43b323804c04d4cea49c169b | \n", "-0.414286 | \n", "-0.435714 | \n", "0.325242 | \n", "-0.754286 | \n", "-0.685714 | \n", "-0.164286 | \n", "-0.147143 | \n", "5.657316 | \n", "... | \n", "2247.347025 | \n", "2681.501259 | \n", "21.624614 | \n", "22.111993 | \n", "7.952462 | \n", "14.197890 | \n", "17.080865 | \n", "26.655742 | \n", "30.708428 | \n", "TrialState.COMPLETE | \n", "
| 19 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "-0.157143 | \n", "-0.185714 | \n", "0.226779 | \n", "-0.431429 | \n", "-0.328571 | \n", "-0.014286 | \n", "0.020000 | \n", "5.974832 | \n", "... | \n", "2902.306123 | \n", "3526.513999 | \n", "17.027383 | \n", "21.682156 | \n", "15.988286 | \n", "9.110958 | \n", "11.100846 | \n", "27.608693 | \n", "40.770037 | \n", "TrialState.COMPLETE | \n", "
20 rows × 38 columns
\n", "| \n", " | hash | \n", "pipeline | \n", "SMAPE_mean | \n", "state | \n", "
|---|---|---|---|---|
| 19 | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8.556535 | \n", "TrialState.COMPLETE | \n", "
| 17 | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8.556535 | \n", "TrialState.COMPLETE | \n", "
| 16 | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8.556535 | \n", "TrialState.COMPLETE | \n", "
| 15 | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8.556535 | \n", "TrialState.COMPLETE | \n", "
| 14 | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8.556535 | \n", "TrialState.COMPLETE | \n", "
| 13 | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8.556535 | \n", "TrialState.COMPLETE | \n", "
| 12 | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8.556535 | \n", "TrialState.COMPLETE | \n", "
| 10 | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8.556535 | \n", "TrialState.COMPLETE | \n", "
| 11 | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8.556535 | \n", "TrialState.COMPLETE | \n", "
| 2 | \n", "7c7932114268832a5458acfecfb453fc | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "9.210183 | \n", "TrialState.COMPLETE | \n", "
| 8 | \n", "25dcd8bb095f87a1ffc499fa6a83ef5d | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "9.943658 | \n", "TrialState.COMPLETE | \n", "
| 4 | \n", "e928929f89156d88ef49e28abaf55847 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "9.946866 | \n", "TrialState.COMPLETE | \n", "
| 0 | \n", "f4f02e1d5f60b8f322a4a8a622dd1c1e | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "9.957781 | \n", "TrialState.COMPLETE | \n", "
| 18 | \n", "6f595f4f43b323804c04d4cea49c169b | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "10.061742 | \n", "TrialState.COMPLETE | \n", "
| 1 | \n", "3d7b7af16d71a36f3b935f69e113e22d | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "10.306909 | \n", "TrialState.COMPLETE | \n", "
| 9 | \n", "3f1ca1759261598081fa3bb2f32fe0ac | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "10.554444 | \n", "TrialState.COMPLETE | \n", "
| 5 | \n", "3b4311d41fcaab7307235ea23b6d4599 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "10.756703 | \n", "TrialState.COMPLETE | \n", "
| 6 | \n", "74065ebc11c81bed6a9819d026c7cd84 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "10.917164 | \n", "TrialState.COMPLETE | \n", "
| 3 | \n", "b7ac5f7fcf9c8959626befe263a9d561 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "11.255320 | \n", "TrialState.COMPLETE | \n", "
| 7 | \n", "b0d0420255c6117045f8254bf8f377a0 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "11.478760 | \n", "TrialState.COMPLETE | \n", "
| \n", " | hash | \n", "pipeline | \n", "SMAPE_mean | \n", "state | \n", "
|---|---|---|---|---|
| 19 | \n", "8363309e454e72993f86f10c7fc7c137 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "8.556535 | \n", "TrialState.COMPLETE | \n", "
| 2 | \n", "7c7932114268832a5458acfecfb453fc | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "9.210183 | \n", "TrialState.COMPLETE | \n", "
| 8 | \n", "25dcd8bb095f87a1ffc499fa6a83ef5d | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "9.943658 | \n", "TrialState.COMPLETE | \n", "
| 4 | \n", "e928929f89156d88ef49e28abaf55847 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "9.946866 | \n", "TrialState.COMPLETE | \n", "
| 0 | \n", "f4f02e1d5f60b8f322a4a8a622dd1c1e | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "9.957781 | \n", "TrialState.COMPLETE | \n", "
| 18 | \n", "6f595f4f43b323804c04d4cea49c169b | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "10.061742 | \n", "TrialState.COMPLETE | \n", "
| 1 | \n", "3d7b7af16d71a36f3b935f69e113e22d | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "10.306909 | \n", "TrialState.COMPLETE | \n", "
| 9 | \n", "3f1ca1759261598081fa3bb2f32fe0ac | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "10.554444 | \n", "TrialState.COMPLETE | \n", "
| 5 | \n", "3b4311d41fcaab7307235ea23b6d4599 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "10.756703 | \n", "TrialState.COMPLETE | \n", "
| 6 | \n", "74065ebc11c81bed6a9819d026c7cd84 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "10.917164 | \n", "TrialState.COMPLETE | \n", "
| 3 | \n", "b7ac5f7fcf9c8959626befe263a9d561 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "11.255320 | \n", "TrialState.COMPLETE | \n", "
| 7 | \n", "b0d0420255c6117045f8254bf8f377a0 | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "11.478760 | \n", "TrialState.COMPLETE | \n", "
| \n", " | hash | \n", "pipeline | \n", "SMAPE_mean | \n", "state | \n", "study | \n", "
|---|---|---|---|---|---|
| 15 | \n", "af8088ac0abfde46e93a8dbb407a2117 | \n", "Pipeline(model = CatBoostPerSegmentModel(itera... | \n", "5.057438 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 2 | \n", "d8215d95e2c6c9a4b4fdacf3fa77dddc | \n", "Pipeline(model = NaiveModel(lag = 7, ), transf... | \n", "5.164436 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 14 | \n", "8f640faabcac0552153ca19337179f3b | \n", "Pipeline(model = HoltWintersModel(trend = 'add... | \n", "5.931951 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 10 | \n", "d6a44adb551f1aec09ef37c14aed260f | \n", "Pipeline(model = SeasonalMovingAverageModel(wi... | \n", "6.197182 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 16 | \n", "16eb77200eb2fd5dc1f6f2a5067884cd | \n", "Pipeline(model = HoltWintersModel(trend = 'add... | \n", "6.347734 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 13 | \n", "4c07749e913403906cd033e4882fc4f9 | \n", "Pipeline(model = SeasonalMovingAverageModel(wi... | \n", "6.529721 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 5 | \n", "6e2eb71d033b6d0607f5b6d0a7596ce9 | \n", "Pipeline(model = ProphetModel(growth = 'linear... | \n", "7.799984 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 17 | \n", "6bb58e7ce09eab00448d5732240ec2ec | \n", "Pipeline(model = CatBoostMultiSegmentModel(ite... | \n", "7.814187 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 18 | \n", "a640ddfb767ea0cbf31751ddda6e36ee | \n", "Pipeline(model = CatBoostMultiSegmentModel(ite... | \n", "7.816528 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 9 | \n", "cfeb21bcf2e922a390ade8be9d845e0d | \n", "Pipeline(model = ProphetModel(growth = 'linear... | \n", "7.893421 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 3 | \n", "a5e036978ef9cc9f297c9eb2c280af05 | \n", "Pipeline(model = AutoARIMAModel(), transforms ... | \n", "8.297048 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 0 | \n", "2e36e0b9cb67a43bb1bf96fa2ccf718f | \n", "Pipeline(model = LinearMultiSegmentModel(fit_i... | \n", "9.205423 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 4 | \n", "8b9f5fa09754a80f17380dec2b998f1d | \n", "Pipeline(model = LinearPerSegmentModel(fit_int... | \n", "10.997462 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 1 | \n", "d62c0579459d4a1b88aea8ed6effdf4e | \n", "Pipeline(model = MovingAverageModel(window = 1... | \n", "11.317256 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 6 | \n", "5916e5b653295271c79caae490618ee9 | \n", "Pipeline(model = MovingAverageModel(window = 2... | \n", "12.028916 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 11 | \n", "5a91b6c8acc2c461913df44fd1429375 | \n", "Pipeline(model = ElasticPerSegmentModel(alpha ... | \n", "12.213320 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 7 | \n", "403b3e18012af5ff9815b408f5c2e47d | \n", "Pipeline(model = MovingAverageModel(window = 4... | \n", "12.243011 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 19 | \n", "6cf8605e6c513053ac4f5203e330c59d | \n", "Pipeline(model = HoltWintersModel(trend = None... | \n", "15.473118 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 12 | \n", "53e90ae4cf7f1f71e6396107549c25ef | \n", "Pipeline(model = NaiveModel(lag = 1, ), transf... | \n", "19.361078 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 8 | \n", "90b31b54cb8c01867be05a3320852682 | \n", "Pipeline(model = ElasticMultiSegmentModel(alph... | \n", "35.971289 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| \n", " | hash | \n", "pipeline | \n", "SMAPE_mean | \n", "state | \n", "study | \n", "
|---|---|---|---|---|---|
| 56 | \n", "419fc80cf634ba0888c4f899f666ad45 | \n", "Pipeline(model = HoltWintersModel(trend = 'mul... | \n", "4.769471 | \n", "TrialState.COMPLETE | \n", "tuning/8f640faabcac0552153ca19337179f3b | \n", "
| 89 | \n", "731ccb72a473bec81789b7f186001ddd | \n", "Pipeline(model = CatBoostPerSegmentModel(itera... | \n", "4.899715 | \n", "TrialState.COMPLETE | \n", "tuning/af8088ac0abfde46e93a8dbb407a2117 | \n", "
| 97 | \n", "9c302769456b4adb9143f11c582f7264 | \n", "Pipeline(model = CatBoostPerSegmentModel(itera... | \n", "4.927197 | \n", "TrialState.COMPLETE | \n", "tuning/af8088ac0abfde46e93a8dbb407a2117 | \n", "
| 88 | \n", "182c748af70287ab3a12bf32c03320f5 | \n", "Pipeline(model = CatBoostPerSegmentModel(itera... | \n", "4.941247 | \n", "TrialState.COMPLETE | \n", "tuning/af8088ac0abfde46e93a8dbb407a2117 | \n", "
| 96 | \n", "4f426335c0eb00d847d9dd1e0a421415 | \n", "Pipeline(model = CatBoostPerSegmentModel(itera... | \n", "4.977773 | \n", "TrialState.COMPLETE | \n", "tuning/af8088ac0abfde46e93a8dbb407a2117 | \n", "
| 98 | \n", "2cafd0750f191e7ab2d4160da50a7c64 | \n", "Pipeline(model = CatBoostPerSegmentModel(itera... | \n", "5.056993 | \n", "TrialState.COMPLETE | \n", "tuning/af8088ac0abfde46e93a8dbb407a2117 | \n", "
| 15 | \n", "af8088ac0abfde46e93a8dbb407a2117 | \n", "Pipeline(model = CatBoostPerSegmentModel(itera... | \n", "5.057438 | \n", "TrialState.COMPLETE | \n", "pool | \n", "
| 75 | \n", "382825866425cac211691205a9537c95 | \n", "Pipeline(model = CatBoostPerSegmentModel(itera... | \n", "5.081609 | \n", "TrialState.COMPLETE | \n", "tuning/af8088ac0abfde46e93a8dbb407a2117 | \n", "
| 95 | \n", "c2a8d498fe35873d060e173e1af042d5 | \n", "Pipeline(model = CatBoostPerSegmentModel(itera... | \n", "5.117583 | \n", "TrialState.COMPLETE | \n", "tuning/af8088ac0abfde46e93a8dbb407a2117 | \n", "
| 91 | \n", "035f8e28180bc7491a30b3d0d67060c9 | \n", "Pipeline(model = CatBoostPerSegmentModel(itera... | \n", "5.135956 | \n", "TrialState.COMPLETE | \n", "tuning/af8088ac0abfde46e93a8dbb407a2117 | \n", "