NeuralForecast

https://nixtlaverse.nixtla.io/neuralforecast/
Admin
NeuralForecast is a user-friendly Python library offering state-of-the-art neural forecasting models...
Tokens:235,862
Snippets:1,967
Trust Score:9
Update:4 months ago
Show doc for...
Tokens:
Raw
### Install datasetsforecast Library

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/getting-started/datarequirements.html

Installs the datasetsforecast library, which is used in the following examples for loading data.

```python
%%capture
!pip install datasetsforecast
```

--------------------------------

### Install NeuralForecast and DatasetsForecast Libraries

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/tutorials/longhorizon_probabilistic.html

Installs the necessary libraries for running the examples. This is a prerequisite for the subsequent code.

```python
%%capture
!pip install neuralforecast datasetsforecast
```

--------------------------------

### HINT Usage Example: Setup and Data Loading

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.hint.html

This snippet demonstrates the initial setup for using the HINT model, including importing necessary libraries, defining an auxiliary sorting function, and loading the TourismLarge hierarchical dataset.

```python
import matplotlib.pyplot as plt

from neuralforecast.losses.pytorch import GMM, sCRPS
from datasetsforecast.hierarchical import HierarchicalData

# Auxiliary sorting
def sort_df_hier(Y_df, S_df):
    # NeuralForecast core, sorts unique_id lexicographically
    # by default, this class matches S_df and Y_hat_df order.    
    Y_df.unique_id = Y_df.unique_id.astype('category')
    Y_df.unique_id = Y_df.unique_id.cat.set_categories(S_df.index)
    Y_df = Y_df.sort_values(by=['unique_id', 'ds'])
    return Y_df

# Load TourismSmall dataset
horizon = 12
Y_df, S_df, tags = HierarchicalData.load('./data', 'TourismLarge')
Y_df['ds'] = pd.to_datetime(Y_df['ds'])
Y_df = sort_df_hier(Y_df, S_df)
level = [80,90]

# Instantiate HINT

```

--------------------------------

### Setup for NeuralForecast Simulation

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/tutorials/simulation.html

Imports necessary libraries and configures logging and warnings for NeuralForecast simulation. Ensure PyTorch is installed with appropriate settings.

```python
import logging
import warnings

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import torch

warnings.filterwarnings("ignore")
logging.getLogger("pytorch_lightning").setLevel(logging.ERROR)
torch.set_float32_matmul_precision("high")
```

--------------------------------

### NeuralForecast Usage Example Setup

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.tft.html

Imports necessary libraries for using the NeuralForecast library, including plotting and data manipulation tools.

```python
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd

from neuralforecast import NeuralForecast
```

--------------------------------

### Install NeuralForecast and Hyperopt

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/hyperparameter_tuning.html

Installs the necessary libraries for NeuralForecast and hyperparameter optimization.

```python
%%capture
!pip install neuralforecast hyperopt
```

--------------------------------

### Install NeuralForecast and Dependencies

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/tutorials/longhorizon_nhits.html

Installs the NeuralForecast library along with datasetsforecast and utilsforecast. This is the initial step for using the library's functionalities.

```python
%%capture
!pip install neuralforecast datasetsforecast utilsforecast
```

--------------------------------

### Install NeuralForecast with Spark Extra

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/getting-started/installation.html

Install NeuralForecast with the 'spark' extra for distributed training capabilities.

```shell
pip install neuralforecast[spark]
```

--------------------------------

### Install Nixtla Libraries

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/tutorials/comparing_methods.html

Installs the necessary libraries for statistical, machine learning, and neural forecasting.

```python
%%capture
!pip install statsforecast mlforecast neuralforecast pyarrow
```

--------------------------------

### Install Necessary Packages

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/tutorials/intermittent_data.html

Installs the required libraries for NeuralForecast, StatsForecast, and data handling. It's recommended to use GPU for faster computation.

```python
%%capture
!pip install statsforecast s3fs fastparquet neuralforecast
```

--------------------------------

### Install NeuralForecast

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/save_load_models.html

Installs the NeuralForecast library. This is a prerequisite for using the library's functionalities.

```python
%%capture
!pip install neuralforecast
```

--------------------------------

### Install NeuralForecast in Dev Mode

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/getting-started/introduction.html

Clone the repository and install NeuralForecast in editable mode for real-time modifications.

```bash
git clone https://github.com/Nixtla/neuralforecast.git
cd neuralforecast
pip install -e .
```

--------------------------------

### Custom LightningDataModule Example

Source: https://nixtlaverse.nixtla.io/neuralforecast/tsdataset.html

Demonstrates how to create a custom LightningDataModule by subclassing L.LightningDataModule. This example includes methods for data preparation, setup, creating data loaders for training, validation, and testing, and handling exceptions and teardown.

```python
import lightning.pytorch as L
import torch.utils.data as data
from pytorch_lightning.demos.boring_classes import RandomDataset

class MyDataModule(L.LightningDataModule):
    def prepare_data(self):
        # download, IO, etc. Useful with shared filesystems
        # only called on 1 GPU/TPU in distributed
        ...

    def setup(self, stage):
        # make assignments here (val/train/test split)
        # called on every process in DDP
        dataset = RandomDataset(1, 100)
        self.train, self.val, self.test = data.random_split(
            dataset, [80, 10, 10], generator=torch.Generator().manual_seed(42)
        )

    def train_dataloader(self):
        return data.DataLoader(self.train)

    def val_dataloader(self):
        return data.DataLoader(self.val)

    def test_dataloader(self):
        return data.DataLoader(self.test)

    def on_exception(self, exception):
        # clean up state after the trainer faced an exception
        ...

    def teardown(self):
        # clean up state after the trainer stops, delete files...
        # called on every process in DDP
        ...*
```

--------------------------------

### Install NeuralForecast with Pip

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/getting-started/installation.html

Use this command to install the released version of NeuralForecast from PyPI.

```shell
pip install neuralforecast
```

--------------------------------

### Install Neuralforecast and Hyperopt

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/time_series_scaling.html

Installs the necessary libraries for neural forecasting experiments. The `%%capture` magic command suppresses output.

```python
%%capture
!pip install neuralforecast
!pip install hyperopt
```

--------------------------------

### Install NeuralForecast in Editable Mode (CPU)

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/getting-started/installation.html

Install NeuralForecast in editable mode with the development dependencies and CPU PyTorch backend using 'uv'.

```bash
$ uv pip install -e ".[dev]" --torch-backend cpu
```

--------------------------------

### Install NeuralForecast with AWS Extra

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/getting-started/installation.html

Install NeuralForecast with the 'aws' extra for saving and loading models from S3.

```shell
pip install neuralforecast[aws]
```

--------------------------------

### RMoK Model Initialization and Prediction Example

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.rmok.html

Example demonstrating how to initialize, train, and predict with the RMoK model using the AirPassengers dataset. Includes plotting of the forecasts.

```python
import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
from neuralforecast.models import RMoK
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic
from neuralforecast.losses.pytorch import MSE

Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

model = RMoK(h=12,
             input_size=24,
             n_series=2,
             taylor_order=3,
             jacobi_degree=6,
             wavelet_function='mexican_hat',
             dropout=0.1,
             revin_affine=True,
             loss=MSE(),
             valid_loss=MAE(),
             early_stop_patience_steps=3,
             batch_size=32)

fcst = NeuralForecast(models=[model], freq='ME')
fcst.fit(df=Y_train_df, static_df=AirPassengersStatic, val_size=12)
forecasts = fcst.predict(futr_df=Y_test_df)

# Plot predictions
fig, ax = plt.subplots(1, 1, figsize = (20, 7))
Y_hat_df = forecasts.reset_index(drop=False).drop(columns=['unique_id','ds'])
plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)
plot_df = pd.concat([Y_train_df, plot_df])

plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
plt.plot(plot_df['ds'], plot_df['RMoK'], c='blue', label='Forecast')
ax.set_title('AirPassengers Forecast', fontsize=22)
ax.set_ylabel('Monthly Passengers', fontsize=20)
ax.set_xlabel('Year', fontsize=20)
ax.legend(prop={'size': 15})
ax.grid()

```

--------------------------------

### MLPMultivariate Model Training and Prediction Example

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.mlpmultivariate.html

This example demonstrates how to initialize, train, and predict using the `MLPMultivariate` model. It includes data preparation, model configuration, fitting, and plotting the forecasts.

```python
import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
from neuralforecast.models import MLPMultivariate
from neuralforecast.losses.pytorch import MAE
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic

Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]] # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

model = MLPMultivariate(h=12, 
            input_size=24,
            n_series=2,
            stat_exog_list=['airline1'],
            futr_exog_list=['trend'],            
            loss = MAE(),
            scaler_type='robust',
            learning_rate=1e-3,
            stat_exog_list=['airline1'],
            max_steps=200,
            val_check_steps=10,
            early_stop_patience_steps=2)

fcst = NeuralForecast(
    models=[model],
    freq='ME'
)
fcst.fit(df=Y_train_df, static_df=AirPassengersStatic, val_size=12)
forecasts = fcst.predict(futr_df=Y_test_df)

# Plot predictions
Y_hat_df = forecasts.reset_index(drop=False).drop(columns=['unique_id','ds'])
plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)
plot_df = pd.concat([Y_train_df, plot_df])

plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
plt.plot(plot_df['ds'], plot_df['MLPMultivariate'], c='blue', label='median')
plt.grid()
plt.legend()
plt.plot()
```

--------------------------------

### Install ML Libraries

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/tutorials/comparing_methods.html

Installs LightGBM and XGBoost, two popular gradient boosting libraries often used with MLForecast for time series modeling.

```python
!pip install lightgbm xgboost
```

--------------------------------

### TimeXer Model Training and Prediction Example

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.timexer.html

This example demonstrates how to initialize, train, and use the TimeXer model for forecasting. It includes data preparation, model configuration, and plotting the predictions against the actual values.

```python
import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
from neuralforecast.models import TimeXer
from neuralforecast.losses.pytorch import MAE, MSE
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic, augment_calendar_df

AirPassengersPanel, calendar_cols = augment_calendar_df(df=AirPassengersPanel, freq='M')

Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]] # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

model = TimeXer(h=12,
                input_size=24,
                n_series=2,
                stat_exog_list=['airline1'],
                patch_len=12,
                hidden_size=128,
                n_heads=16,
                e_layers=2,
                d_ff=256,
                factor=1,
                dropout=0.1,
                use_norm=True,
                loss=MSE(),
                valid_loss=MAE(),
                early_stop_patience_steps=3,
                batch_size=32)

fcst = NeuralForecast(models=[model], freq='ME')
fcst.fit(df=Y_train_df, static_df=AirPassengersStatic, val_size=12)
forecasts = fcst.predict(futr_df=Y_test_df)

# Plot predictions
fig, ax = plt.subplots(1, 1, figsize = (20, 7))
Y_hat_df = forecasts.reset_index(drop=False).drop(columns=['unique_id','ds'])
plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)
plot_df = pd.concat([Y_train_df, plot_df])

plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
plt.plot(plot_df['ds'], plot_df['TimeXer'], c='blue', label='Forecast')
ax.set_title('AirPassengers Forecast', fontsize=22)
ax.set_ylabel('Monthly Passengers', fontsize=20)
ax.set_xlabel('Year', fontsize=20)
ax.legend(prop={'size': 15})
ax.grid()
```

--------------------------------

### BiTCN Usage Example

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.bitcn.html

Demonstrates how to use the BiTCN model for time series forecasting. This example includes data preparation, model initialization with specific parameters like input size, loss function, and exogenous variables, model fitting, prediction, and plotting of results including quantiles.

```python
import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
from neuralforecast.losses.pytorch import GMM
from neuralforecast.models import BiTCN
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic

Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]] # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

fcst = NeuralForecast(
    models=[
            BiTCN(h=12,
                input_size=24,
                loss=GMM(n_components=7, level=[80,90]),
                max_steps=100,
                scaler_type='standard',
                futr_exog_list=['y_[lag12]'],
                hist_exog_list=None,
                stat_exog_list=['airline1'],
                windows_batch_size=2048,
                val_check_steps=10,
                early_stop_patience_steps=-1,
                ),
    ],
    freq='ME'
)
fcst.fit(df=Y_train_df, static_df=AirPassengersStatic)
forecasts = fcst.predict(futr_df=Y_test_df)

# Plot quantile predictions
Y_hat_df = forecasts.reset_index(drop=False).drop(columns=['unique_id','ds'])
plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)
plot_df = pd.concat([Y_train_df, plot_df])

plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
plt.plot(plot_df['ds'], plot_df['BiTCN-median'], c='blue', label='median')
plt.fill_between(x=plot_df['ds'][-12:], 
                 y1=plot_df['BiTCN-lo-90'][-12:].values,
                 y2=plot_df['BiTCN-hi-90'][-12:].values,
                 alpha=0.4, label='level 90')
plt.legend()
plt.grid()
```

--------------------------------

### SOFTS Model Usage Example

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.softs.html

Example demonstrating how to initialize, fit, and predict using the SOFTS model with the NeuralForecast framework. Includes data preparation and plotting of results.

```python
import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
from neuralforecast.models import SOFTS
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic
from neuralforecast.losses.pytorch import MASE
Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

model = SOFTS(h=12,
              input_size=24,
              n_series=2,
              hidden_size=256,
              d_core=256,
              e_layers=2,
              d_ff=64,
              dropout=0.1,
              use_norm=True,
              loss=MASE(seasonality=4),
              early_stop_patience_steps=3,
              batch_size=32)

fcst = NeuralForecast(models=[model], freq='ME')
fcst.fit(df=Y_train_df, static_df=AirPassengersStatic, val_size=12)
forecasts = fcst.predict(futr_df=Y_test_df)

# Plot predictions
fig, ax = plt.subplots(1, 1, figsize = (20, 7))
Y_hat_df = forecasts.reset_index(drop=False).drop(columns=['unique_id','ds'])
plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)
plot_df = pd.concat([Y_train_df, plot_df])

plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
plt.plot(plot_df['ds'], plot_df['SOFTS'], c='blue', label='Forecast')
ax.set_title('AirPassengers Forecast', fontsize=22)
ax.set_ylabel('Monthly Passengers', fontsize=20)
ax.set_xlabel('Year', fontsize=20)
ax.legend(prop={'size': 15})
ax.grid()
```

--------------------------------

### TCN Model Prediction and Visualization Example

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.tcn.html

This example demonstrates how to initialize a TCN model, train it on historical data, make future predictions, and visualize the results along with confidence intervals.

```python
import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
from neuralforecast.models import TCN
from neuralforecast.losses.pytorch import  DistributionLoss
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic

Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]] # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

fcst = NeuralForecast(
    models=[TCN(h=12,
                input_size=-1,
                loss=DistributionLoss(distribution='Normal', level=[80, 90]),
                learning_rate=5e-4,
                kernel_size=2,
                dilations=[1,2,4,8,16],
                encoder_hidden_size=128,
                context_size=10,
                decoder_hidden_size=128,
                decoder_layers=2,
                max_steps=500,
                scaler_type='robust',
                futr_exog_list=['y_[lag12]'],
                hist_exog_list=None,
                stat_exog_list=['airline1'],
                )
    ],
    freq='ME'
)
fcst.fit(df=Y_train_df, static_df=AirPassengersStatic)
forecasts = fcst.predict(futr_df=Y_test_df)

# Plot quantile predictions
Y_hat_df = forecasts.reset_index(drop=False).drop(columns=['unique_id','ds'])
plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)
plot_df = pd.concat([Y_train_df, plot_df])

plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
plt.plot(plot_df['ds'], plot_df['TCN-median'], c='blue', label='median')
plt.fill_between(x=plot_df['ds'][-12:], 
                 y1=plot_df['TCN-lo-90'][-12:].values,
                 y2=plot_df['TCN-hi-90'][-12:].values,
                 alpha=0.4, label='level 90')
plt.legend()
plt.grid()
plt.plot()
```

--------------------------------

### Install NeuralForecast in Editable Mode (CUDA)

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/getting-started/installation.html

Install NeuralForecast in editable mode with the development dependencies and CUDA 11.8 PyTorch backend using 'uv'.

```bash
$ uv pip install -e ".[dev]" --torch-backend cu118
```

--------------------------------

### XLinear Model Training and Prediction Example

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.xlinear.html

This example demonstrates how to train and predict using the XLinear model. It includes data preparation, model initialization with various parameters, fitting the model to training data, and generating forecasts for test data. The snippet also includes plotting the predictions against the actual values.

```python
import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
from neuralforecast.models import XLinear
from neuralforecast.losses.pytorch import MAE
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic

Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]] # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

model = XLinear(h=12, 
            input_size=24,
            n_series=2,
            stat_exog_list=['airline1'],
            hist_exog_list=["y_[lag12]"]
            futr_exog_list=['trend'],            
            loss = MAE(),
            scaler_type='robust',
            learning_rate=1e-3,
            max_steps=200,
            val_check_steps=10,
            early_stop_patience_steps=2)

fcst = NeuralForecast(
    models=[model],
    freq='ME'
)
fcst.fit(df=Y_train_df, static_df=AirPassengersStatic, val_size=12)
forecasts = fcst.predict(futr_df=Y_test_df)

# Plot predictions
Y_hat_df = forecasts.reset_index(drop=False).drop(columns=['unique_id','ds'])
plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)
plot_df = pd.concat([Y_train_df, plot_df])

plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
plt.plot(plot_df['ds'], plot_df['XLinear'], c='blue', label='median')
plt.grid()
plt.legend()
plt.plot()
```

--------------------------------

### iTransformer Model Training and Prediction Example

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.itransformer.html

This example demonstrates how to initialize an iTransformer model, train it using NeuralForecast, and then generate predictions on a test dataset. It also includes plotting the forecast against the actual values.

```python
import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
from neuralforecast.models import iTransformer
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic
from neuralforecast.losses.pytorch import MSE

Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

model = iTransformer(h=12,
                     input_size=24,
                     n_series=2,
                     hidden_size=128,
                     n_heads=2,
                     e_layers=2,
                     d_layers=1,
                     d_ff=4,
                     factor=1,
                     dropout=0.1,
                     use_norm=True,
                     loss=MSE(),
                     valid_loss=MAE(),
                     early_stop_patience_steps=3,
                     batch_size=32,
                     max_steps=100)

fcst = NeuralForecast(models=[model], freq='ME')
fcst.fit(df=Y_train_df, static_df=AirPassengersStatic, val_size=12)
forecasts = fcst.predict(futr_df=Y_test_df)

# Plot predictions
fig, ax = plt.subplots(1, 1, figsize = (20, 7))
Y_hat_df = forecasts.reset_index(drop=False).drop(columns=['unique_id','ds'])
plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)
plot_df = pd.concat([Y_train_df, plot_df])

plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
plt.plot(plot_df['ds'], plot_df['iTransformer'], c='blue', label='Forecast')
ax.set_title('AirPassengers Forecast', fontsize=22)
ax.set_ylabel('Monthly Passengers', fontsize=20)
ax.set_xlabel('Year', fontsize=20)
ax.legend(prop={'size': 15})
ax.grid()
```

--------------------------------

### BaseAuto Initialization and Fit with Optuna

Source: https://nixtlaverse.nixtla.io/neuralforecast/common.base_auto.html

Example of initializing BaseAuto with an Optuna configuration function and sampler, then fitting it to a dataset. It verifies the results object type and prediction accuracy.

```python
auto2 = BaseAuto(h=12, loss=MAE(), valid_loss=MSE(), cls_model=MLP, config=config_f, search_alg=optuna.samplers.RandomSampler(), num_samples=2, backend='optuna', callbacks=[OptunaLogLossesCallback()])
auto2.fit(dataset=dataset)
assert isinstance(auto2.results, optuna.Study)
y_hat2 = auto2.predict(dataset=dataset)
assert mae(Y_test_df['y'].values, y_hat2[:, 0]) < 200

```

--------------------------------

### BaseAuto Initialization and Fit with Ray Tune

Source: https://nixtlaverse.nixtla.io/neuralforecast/common.base_auto.html

Example of initializing BaseAuto with a configuration for Ray Tune and fitting it to a dataset. It includes a custom callback for logging.

```python
config = {
    "hidden_size": tune.choice([512]),
    "num_layers": tune.choice([3, 4]),
    "input_size": 12,
    "max_steps": 10,
    "val_check_steps": 5
}
auto = BaseAuto(h=12, loss=MAE(), valid_loss=MSE(), cls_model=MLP, config=config, num_samples=2, cpus=1, gpus=0, callbacks=[RayLogLossesCallback()])
auto.fit(dataset=dataset)
y_hat = auto.predict(dataset=dataset)
assert mae(Y_test_df['y'].values, y_hat[:, 0]) < 200

```

--------------------------------

### Setup Imports and Warnings

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/tutorials/weighting_timesteps.html

Imports necessary libraries for NeuralForecast, data manipulation, evaluation, and plotting. It also configures warnings and logging levels.

```python
import logging
import warnings

import numpy as np

from utilsforecast.evaluation import evaluate
from utilsforecast.losses import mae
from utilsforecast.plotting import plot_series

from neuralforecast import NeuralForecast
from neuralforecast.models import NHITS

warnings.filterwarnings("ignore")
logging.getLogger("pytorch_lightning").setLevel(logging.ERROR)
```

--------------------------------

### Install NeuralForecast with Conda

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/getting-started/installation.html

Use this command to install the released version of NeuralForecast from the conda-forge channel.

```shell
conda install -c conda-forge neuralforecast
```

--------------------------------

### Install Seaborn for Visualization

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/tutorials/comparing_methods.html

Installs the seaborn library, which is used for creating statistical visualizations, including error distributions.

```python
%%capture
!pip install seaborn
```

--------------------------------

### Configure HINT Model Parameters

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/tutorials/hierarchical_forecasting.html

Sets up the HINT model, including the forecast horizon, input size, loss function (GMM with quantiles), exogenous variables, and training parameters like max steps and early stopping. It also defines the base NHITS model and the reconciliation strategy.

```python
# Horizon and quantiles
level = np.arange(0, 100, 2)
qs = [[50-lv/2, 50+lv/2] if lv!=0 else [50] for lv in level]
quantiles = np.sort(np.concatenate(qs)/100)

# HINT := BaseNetwork + Distribution + Reconciliation
nhits = NHITS(h=horizon,
              input_size=24,
              loss=GMM(n_components=10, quantiles=quantiles),
              hist_exog_list=['month'],
              max_steps=2000,
              early_stop_patience_steps=10,
              val_check_steps=50,
              scaler_type='robust',
              learning_rate=1e-3,
              valid_loss=sCRPS(quantiles=quantiles))

model = HINT(h=horizon, S=S_df.drop(columns='unique_id').values,
             model=nhits,  reconciliation='BottomUp')
```

--------------------------------

### Install Packages

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/tutorials/hierarchical_forecasting.html

Installs the required packages for hierarchical forecasting: datasetsforecast, hierarchicalforecast, neuralforecast, and statsforecast. This command is typically run in a notebook environment.

```python
%%capture
!pip install datasetsforecast hierarchicalforecast neuralforecast statsforecast
```

--------------------------------

### Prepare Data and Initialize Models

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/getting-started/introduction.html

Load the AirPassengers dataset, split it into training and testing sets, and initialize NBEATS and NHITS models with specified parameters.

```python
# Split data and declare panel dataset
Y_df = AirPassengersDF
Y_train_df = Y_df[Y_df.ds<='1959-12-31'] # 132 train
Y_test_df = Y_df[Y_df.ds>'1959-12-31'] # 12 test

# Fit and predict with NBEATS and NHITS models
horizon = len(Y_test_df)
models = [NBEATS(input_size=2 * horizon, h=horizon, max_steps=100, enable_progress_bar=False),
          NHITS(input_size=2 * horizon, h=horizon, max_steps=100, enable_progress_bar=False)]
nf = NeuralForecast(models=models, freq='ME')
nf.fit(df=Y_train_df)
Y_hat_df = nf.predict()
```

--------------------------------

### DilatedRNN Usage Example

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.dilated_rnn.html

Demonstrates how to use the DilatedRNN model for time series forecasting. This example includes data preparation, model initialization, training, prediction, and visualization of results.

```python
import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
from neuralforecast.models import DilatedRNN
from neuralforecast.losses.pytorch import DistributionLoss
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic

Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]] # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

fcst = NeuralForecast(
    models=[DilatedRNN(h=12,
                       input_size=-1,
                       loss=DistributionLoss(distribution='Normal', level=[80, 90]),
                       scaler_type='robust',
                       encoder_hidden_size=100,
                       max_steps=200,
                       futr_exog_list=['y_[lag12]'],
                       hist_exog_list=None,
                       stat_exog_list=['airline1'],
    )
    ],
    freq='ME'
)
fcst.fit(df=Y_train_df, static_df=AirPassengersStatic)
forecasts = fcst.predict(futr_df=Y_test_df)

Y_hat_df = forecasts.reset_index(drop=False).drop(columns=['unique_id','ds'])
plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)
plot_df = pd.concat([Y_train_df, plot_df])

plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
plt.plot(plot_df['ds'], plot_df['DilatedRNN-median'], c='blue', label='median')
plt.fill_between(x=plot_df['ds'][-12:], 
                 y1=plot_df['DilatedRNN-lo-90'][-12:].values,
                 y2=plot_df['DilatedRNN-hi-90'][-12:].values,
                 alpha=0.4, label='level 90')
plt.legend()
plt.grid()
plt.plot()
```

--------------------------------

### Instantiate BiTCN Model

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.bitcn.html

This snippet shows how to instantiate the BiTCN model with its required and optional parameters. Ensure you provide the forecast horizon (h) and input size (input_size).

```python
BiTCN(
    h,
    input_size,
    hidden_size=16,
    dropout=0.5,
    futr_exog_list=None,
    hist_exog_list=None,
    stat_exog_list=None,
    exclude_insample_y=False,
    loss=MAE(),
    valid_loss=None,
    max_steps=1000,
    learning_rate=0.001,
    num_lr_decays=-1,
    early_stop_patience_steps=-1,
    val_monitor="ptl/val_loss",
    val_check_steps=100,
    batch_size=32,
    valid_batch_size=None,
    windows_batch_size=1024,
    inference_windows_batch_size=1024,
    start_padding_enabled=False,
    training_data_availability_threshold=0.0,
    step_size=1,
    scaler_type="identity",
    random_seed=1,
    drop_last_loader=False,
    alias=None,
    optimizer=None,
    optimizer_kwargs=None,
    lr_scheduler=None,
    lr_scheduler_kwargs=None,
    dataloader_kwargs=None,
    **trainer_kwargs
)
```

--------------------------------

### DeepNPTS Usage Example

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.deepnpts.html

Example demonstrating how to use the DeepNPTS model for time series forecasting. It includes data preparation, model initialization, fitting, prediction, and plotting the results.

```python
import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
from neuralforecast.models import DeepNPTS
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic

Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]] # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

nf = NeuralForecast(
    models=[DeepNPTS(h=12,
                   input_size=24,
                   stat_exog_list=['airline1'],
                   futr_exog_list=['trend'],
                   max_steps=1000,
                   val_check_steps=10,
                   early_stop_patience_steps=3,
                   scaler_type='robust',
                   enable_progress_bar=True),
    ],
    freq='ME'
)
nf.fit(df=Y_train_df, static_df=AirPassengersStatic, val_size=12)
Y_hat_df = nf.predict(futr_df=Y_test_df)

# Plot quantile predictions
Y_hat_df = Y_hat_df.reset_index(drop=False).drop(columns=['unique_id','ds'])
plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)
plot_df = pd.concat([Y_train_df, plot_df])

plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
plt.plot(plot_df['ds'], plot_df['DeepNPTS'], c='red', label='mean')
plt.grid()
plt.plot()
```

--------------------------------

### Import Optuna Logging

Source: https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/hyperparameter_tuning.html

Import the Optuna library and set logging verbosity to WARNING to suppress training output.

```python
import optuna
```

--------------------------------

### HINT Model Initialization

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.hint.html

Initializes the HINT model with forecast horizon, model architecture, summing matrix, and reconciliation method. Available reconciliation methods include 'BottomUp', 'MinTraceOLS', and 'MinTraceWLS'.

```python
HINT(h, S, model, reconciliation, alias=None)
```

--------------------------------

### KAN Model Usage Example

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.kan.html

This example demonstrates how to use the KAN model for time series forecasting. It includes data preparation, model initialization, fitting, prediction, and plotting the results.

```python
import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
from neuralforecast.models import KAN
from neuralforecast.losses.pytorch import DistributionLoss
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic

Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]] # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

fcst = NeuralForecast(
    models=[
            KAN(h=12,
                input_size=24,
                loss = DistributionLoss(distribution="Normal"),
                max_steps=100,
                scaler_type='standard',
                futr_exog_list=['y_[lag12]'],
                hist_exog_list=None,
                stat_exog_list=['airline1'],
                ),
    ],
    freq='ME'
)
fcst.fit(df=Y_train_df, static_df=AirPassengersStatic)
forecasts = fcst.predict(futr_df=Y_test_df)

# Plot quantile predictions
Y_hat_df = forecasts.reset_index(drop=False).drop(columns=['unique_id','ds'])
plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)
plot_df = pd.concat([Y_train_df, plot_df])

plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
plt.plot(plot_df['ds'], plot_df['KAN-median'], c='blue', label='median')
plt.fill_between(x=plot_df['ds'][-12:], 
                 y1=plot_df['KAN-lo-90'][-12:].values,
                 y2=plot_df['KAN-hi-90'][-12:].values,
                 alpha=0.4, label='level 90')
plt.legend()
plt.grid()

```

--------------------------------

### PatchTST Model Training and Forecasting Example

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.patchtst.html

Demonstrates a complete workflow for training a PatchTST model, generating forecasts, and visualizing the results using the AirPassengers dataset. Includes data preparation, model instantiation with distribution loss, fitting, prediction, and plotting.

```python
import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
from neuralforecast.models import PatchTST
from neuralforecast.losses.pytorch import DistributionLoss
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic, augment_calendar_df

AirPassengersPanel, calendar_cols = augment_calendar_df(df=AirPassengersPanel, freq='M')

Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]] # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

model = PatchTST(h=12,
                 input_size=104,
                 patch_len=24,
                 stride=24,
                 revin=False,
                 hidden_size=16,
                 n_heads=4,
                 scaler_type='robust',
                 loss=DistributionLoss(distribution='StudentT', level=[80, 90]),
                 learning_rate=1e-3,
                 max_steps=500,
                 val_check_steps=50,
                 early_stop_patience_steps=2)

nf = NeuralForecast(
    models=[model],
    freq='ME'
)
nf.fit(df=Y_train_df, static_df=AirPassengersStatic, val_size=12)
forecasts = nf.predict(futr_df=Y_test_df)

Y_hat_df = forecasts.reset_index(drop=False).drop(columns=['unique_id','ds'])
plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)
plot_df = pd.concat([Y_train_df, plot_df])

if model.loss.is_distribution_output:
    plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
    plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
    plt.plot(plot_df['ds'], plot_df['PatchTST-median'], c='blue', label='median')
    plt.fill_between(x=plot_df['ds'][-12:], 
                    y1=plot_df['PatchTST-lo-90'][-12:].values, 
                    y2=plot_df['PatchTST-hi-90'][-12:].values,
                    alpha=0.4, label='level 90')
    plt.grid()
    plt.legend()
    plt.plot()
else:
    plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
    plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
    plt.plot(plot_df['ds'], plot_df['PatchTST'], c='blue', label='Forecast')
    plt.legend()
    plt.grid()
```

--------------------------------

### GRU Model Usage Example

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.gru.html

Example demonstrating how to use the GRU model for time series forecasting. It includes data loading, model initialization, fitting, prediction, and plotting the results.

```python
import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
# from neuralforecast.models import GRU
from neuralforecast.losses.pytorch import DistributionLoss
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic
Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]] # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

fcst = NeuralForecast(
    models=[GRU(h=12, input_size=24,
                loss=DistributionLoss(distribution='Normal', level=[80, 90]),
                scaler_type='robust',
                encoder_n_layers=2,
                encoder_hidden_size=128,
                decoder_hidden_size=128,
                decoder_layers=2,
                max_steps=200,
                futr_exog_list=None,
                hist_exog_list=['y_[lag12]'],
                stat_exog_list=['airline1'],
                )
    ],
    freq='ME'
)
fcst.fit(df=Y_train_df, static_df=AirPassengersStatic)
forecasts = fcst.predict(futr_df=Y_test_df)

Y_hat_df = forecasts.reset_index(drop=False).drop(columns=['unique_id','ds'])
plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)
plot_df = pd.concat([Y_train_df, plot_df])

plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
plt.plot(plot_df['ds'], plot_df['GRU-median'], c='blue', label='median')
plt.fill_between(x=plot_df['ds'][-12:], 
                 y1=plot_df['GRU-lo-90'][-12:].values, 
                 y2=plot_df['GRU-hi-90'][-12:].values,
                 alpha=0.4, label='level 90')
plt.legend()
plt.grid()
plt.plot()
```

--------------------------------

### TiDE Model Prediction and Plotting Example

Source: https://nixtlaverse.nixtla.io/neuralforecast/models.tide.html

Demonstrates how to fit a TiDE model and generate predictions using sample data. Includes plotting the forecasted values against the true values and confidence intervals.

```python
import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
from neuralforecast.models import TiDE
from neuralforecast.losses.pytorch import GMM
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic

Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]] # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

fcst = NeuralForecast(
    models=[
            TiDE(h=12,
                input_size=24,
                loss=GMM(n_components=7, return_params=True, level=[80,90], weighted=True),
                max_steps=100,
                scaler_type='standard',
                futr_exog_list=['y_[lag12]'],
                hist_exog_list=None,
                stat_exog_list=['airline1'],
                ),
    ],
    freq='ME'
)
fcst.fit(df=Y_train_df, static_df=AirPassengersStatic)
forecasts = fcst.predict(futr_df=Y_test_df)

# Plot quantile predictions
Y_hat_df = forecasts.reset_index(drop=False).drop(columns=['unique_id','ds'])
plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)
plot_df = pd.concat([Y_train_df, plot_df])

plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
plt.plot(plot_df['ds'], plot_df['TiDE-median'], c='blue', label='median')
plt.fill_between(x=plot_df['ds'][-12:], 
                 y1=plot_df['TiDE-lo-90'][-12:].values,
                 y2=plot_df['TiDE-hi-90'][-12:].values,
                 alpha=0.4, label='level 90')
plt.legend()
plt.grid()

```
Tokens:
Raw