### Run Quickstart Experiment

Source: https://github.com/mlfoundations/tableshift/blob/main/examples/README.md

Execute a single experiment to test your setup. This is a good starting point for verifying your installation.

```bash
python run_expt.py --experiment adult --model histgbm
```

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

### Run Container to Test Installation

Source: https://github.com/mlfoundations/tableshift/blob/main/docker/README.md

Execute the container with a sample script to verify your Docker setup. Mounts a local tmp directory for output.

```bash
docker run -v $(pwd)/tmp:/tableshift/tmp tableshift --model lightgbm
```

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

### Get ANES Dataset

Source: https://github.com/mlfoundations/tableshift/blob/main/docs/datasets.md

Verify installation by fetching the ANES dataset. The data will be automatically downloaded to the cache if not already present.

```python
from tableshift import get_dataset
dset = get_dataset("anes")
```

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

### Run Docker Container for Testing

Source: https://github.com/mlfoundations/tableshift/blob/main/README.md

Run the Docker container and automatically launch the example script to test your setup. The --model argument specifies the model to use.

```bash
docker run ghcr.io/jpgard/tableshift:latest --model xgb
```

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

### Test Conda Environment Installation

Source: https://github.com/mlfoundations/tableshift/blob/main/README.md

Test the Conda environment installation by running the main training script. Successful execution will show 'training completed!' with an accuracy score.

```bash
python examples/run_expt.py
```

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

### Run Container Interactively

Source: https://github.com/mlfoundations/tableshift/blob/main/docker/README.md

Start the Docker container in interactive mode, providing a bash shell. Mounts a local tmp directory.

```bash
docker run -it --entrypoint=/bin/bash -v $(pwd)/tmp:/tableshift/tmp tableshift
```

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

### Run Distributed Hyperparameter Tuning with Ray

Source: https://github.com/mlfoundations/tableshift/blob/main/examples/README.md

Conduct hyperparameter tuning using Ray. This example demonstrates a toy run with specified experiment, sample count, worker configuration, and model selection.

```bash
ulimit -u 127590 && python scripts/ray_train.py \
	--experiment adult \
	--num_samples 2 \
	--num_workers 1 \
	--cpu_per_worker 4 \
	--use_cached \
	--models xgb
```

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

### Fetch Docker Image

Source: https://github.com/mlfoundations/tableshift/blob/main/README.md

Fetch the latest TableShift Docker image from Docker Hub. This is the recommended method for environment setup.

```bash
docker pull ghcr.io/jpgard/tableshift:latest
```

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

### Get a Standard TableShift Dataset

Source: https://github.com/mlfoundations/tableshift/blob/main/README.md

Load a standard version of a public TableShift dataset using its name. The 'diabetes_readmission' dataset is used as an example.

```python
from tableshift import get_dataset

dataset_name = "diabetes_readmission"
dset = get_dataset(dataset_name)
```

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

### Train Single XGBoost Model

Source: https://github.com/mlfoundations/tableshift/blob/main/examples/README.md

Train a single XGBoost model on the UCI Adult dataset for testing or debugging your setup.

```bash
python run_expt.py --experiment adult --model xgb
```

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

### Train Single Tabular ResNet Model

Source: https://github.com/mlfoundations/tableshift/blob/main/examples/README.md

Train a single Tabular ResNet model on the UCI Adult dataset for testing or debugging your setup.

```bash
python run_expt.py --experiment adult --model resnet
```

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

### Train and Tune CatBoost Model with Optuna

Source: https://github.com/mlfoundations/tableshift/blob/main/examples/README.md

Train and tune a CatBoost model using Optuna, as CatBoost is not officially supported in Ray. This script allows for hyperparameter tuning with specified experiment and sample count.

```bash
python scripts/train_catboost_optuna.py --experiment $EXPERIMENT --use_cached --num_samples 100
```

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

### Build the TableShift Docker Image Locally

Source: https://github.com/mlfoundations/tableshift/blob/main/docker/README.md

Build the Docker image from the local Dockerfile. Ensure you are in the project root directory.

```bash
docker build -t tableshift -f docker/tableshift.dockerfile .
```

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

### Create and Activate Conda Environment

Source: https://github.com/mlfoundations/tableshift/blob/main/README.md

Set up a Conda environment for TableShift by creating an environment from the 'environment.yml' file and activating it.

```bash
conda env create -f environment.yml
conda activate tableshift
```

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

### Cache Data for Experiments

Source: https://github.com/mlfoundations/tableshift/blob/main/examples/README.md

Cache datasets before running experiments. Caching fetches, preprocesses, and splits remote data, writing it to disk for efficient distributed training and handling large datasets.

```bash
python scripts/cache_task.py --experiment $EXPERIMENT
```

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

### Accessing Datasets with TableShift

Source: https://github.com/mlfoundations/tableshift/blob/main/benchmarking_guide.md

Use `get_dataset` to fetch benchmark data. Supports Pandas DataFrames and PyTorch DataLoaders. Ensure the `experiment` name and `cache_dir` are correctly specified.

```python
from tableshift import get_dataset
experiment = "diabetes_readmission"  # can be any TableShift dataset name

dset = get_dataset(experiment, cache_dir)

# For Pandas DataFrames
X, y, _, _ = dset.get_pandas("train")
X_test_id, y_test_id, _, _ = dset.get_pandas("test")
X_test_ood, y_test_ood, _, _ = dset.get_pandas("ood_test")

# PyTorch DataLoader
train_loader = dset.get_dataloader("train", batch_size=1024)
id_test_loader = dset.get_dataloader("test", batch_size=1024)
ood_test_loader = dset.get_dataloader("ood_test", batch_size=1024)
```

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

### Pull the TableShift Docker Image

Source: https://github.com/mlfoundations/tableshift/blob/main/docker/README.md

Use this command to pull the pre-built TableShift container from Docker Hub.

```bash
docker pull
```

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

### Run Docker Container Interactively

Source: https://github.com/mlfoundations/tableshift/blob/main/README.md

Run the Docker container interactively, providing a bash shell for development or exploration within the container.

```bash
docker run -it --entrypoint=/bin/bash ghcr.io/jpgard/tableshift:latest
```

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

### Run TableShift Experiment

Source: https://github.com/mlfoundations/tableshift/blob/main/reproduction_guide.md

Execute a TableShift experiment using the `ray_train.py` script. This command can be used with different model and dataset names.

```bash
python scripts/ray_train.py \
    --models mlp \
    --experiment diabetes_readmission
```

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

### Fetch Dataset as PyTorch DataLoader

Source: https://github.com/mlfoundations/tableshift/blob/main/README.md

Load the training set of a fetched TableShift dataset as a PyTorch DataLoader. This allows for batch processing of data during model training.

```python
# Fetch and use a pytorch DataLoader
train_loader = dset.get_dataloader("train", batch_size=1024)

for X, y, _, _ in train_loader:
    ...
```

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

### Caching Datasets with TableShift

Source: https://github.com/mlfoundations/tableshift/blob/main/benchmarking_guide.md

Cache preprocessed datasets to sharded files for faster loading. Use `to_sharded` to specify shard size and file type. Cached datasets can be loaded using `use_cached=True`.

```python
dset = get_dataset(experiment, cache_dir)
dset.to_sharded(rows_per_shard=4096, file_type="csv")
```

```python
dset = get_dataset(experiment, cache_dir, use_cached=True)
```

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

### Train and Evaluate Scikit-learn Model with TableShift

Source: https://github.com/mlfoundations/tableshift/blob/main/README.md

Load a dataset, train a Gradient Boosting Classifier, and evaluate its accuracy on different splits. Ensure the 'diabetes_readmission' dataset is available.

```python
from tableshift import get_dataset
from sklearn.ensemble import GradientBoostingClassifier

dset = get_dataset("diabetes_readmission")
X_train, y_train, _, _ = dset.get_pandas("train")

# Train
estimator = GradientBoostingClassifier()
trained_estimator = estimator.fit(X_train, y_train)

# Test
for split in ('id_test', 'ood_test'):
    X, y, _, _ = dset.get_pandas(split)
    preds = estimator.predict(X)
    acc = (preds == y).mean()
    print(f'accuracy on split {split} is: {acc:.3f}')
```

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

### Fetch Dataset as Pandas DataFrame

Source: https://github.com/mlfoundations/tableshift/blob/main/README.md

Load the training set of a fetched TableShift dataset as a Pandas DataFrame. The function returns training and testing features and labels.

```python
# Fetch a pandas DataFrame of the training set
X_tr, y_tr, _, _ = dset.get_pandas("train")
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.