### Install hboard (Web Server)

Source: https://context7.com/datacanvasio/hyperboard/llms.txt

Install the web-based visualization server using pip or conda.

```bash
# Install via pip
pip install hboard
```

```bash
# Or install via conda
conda install -c conda-forge hboard
```

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

### Start HyperBoard web server

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard/README.md

Initialize and start the web server to monitor the specified event file.

```python
from hboard.app import WebApp
webapp = WebApp("events.txt")
webapp.start()
```

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

### CLI - Start Web Server

Source: https://context7.com/datacanvasio/hyperboard/llms.txt

Use the `hboard` command-line tool to visualize completed experiments by starting a web server. You can specify the event file path and a custom port.

```bash
# View CLI help
hboard -h
# usage: hboard [-h] {server} ...
# hboard command is used to visualize the experiment
```

```bash
# Start web server to visualize a completed experiment
hboard server --event-file=/path/to/events.json
```

```bash
# Specify a custom port
hboard server --event-file=/path/to/events.json --port=9999
```

```bash
# Example with sample data from repository
hboard server --event-file=hboard/hboard/tests/events_example.json
# Access dashboard at http://localhost:8888
```

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

### Start Visualization Web Server

Source: https://context7.com/datacanvasio/hyperboard/llms.txt

Start the HyperBoard web visualization server. Ensure `event_file` is defined before instantiation.

```python
from hboard.app import WebApp
webapp = WebApp(event_file=event_file, server_port=8888)
webapp.start()
```

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

### Install hboard-widget (Jupyter Integration)

Source: https://context7.com/datacanvasio/hyperboard/llms.txt

Install the Jupyter widget for inline experiment visualization in notebooks.

```bash
# Install via pip
pip install hboard-widget
```

```bash
# Or install via conda
conda install -c conda-forge hboard-widget
```

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

### Install HyperBoard via pip

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard/README.md

Standard installation method using the Python package manager.

```shell
pip install hboard
```

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

### View server logs

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard/README.md

Example output showing the server running status.

```shell
02-24 20:45:58 I hboard.app.py 77 - experiment visualization http server is running at: http://0.0.0.0:8888
```

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

### Initialize hboard-frontend Project

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard-frontend/README.md

Navigates into the hboard-frontend directory, installs project dependencies using yarn, and builds the project using webpack. Ensure you have already cloned the repository and installed global packages.

```bash
cd HyperBoard/hboard-frontend
yarn
webpack
```

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

### Install hboard-widget with pip

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard-widget/README.md

Use this command to install the hboard-widget package using pip.

```shell
pip install hboard-widget
```

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

### Install HyperBoard via conda

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard/README.md

Alternative installation method using the conda package manager.

```shell
conda install -c conda-forge hboard
```

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

### Install Global Packages

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard-frontend/README.md

Installs webpack, webpack-cli, and yarn globally using npm. Ensure Node.js v14.15.0+ is installed.

```bash
npm install -g webpack webpack-cli yarn
```

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

### Programmatically Create Experiment Events

Source: https://context7.com/datacanvasio/hyperboard/llms.txt

Demonstrates how to programmatically create and write experiment events to an NDJSON file using Python. Includes functions for writing events and examples for experiment start, step progression, and trial results.

```python
import json

def write_event(event_file, event_type, payload):
    """Append an event to the experiment event file."""
    event = {"type": event_type, "payload": payload}
    with open(event_file, 'a', newline='\n') as f:
        f.write(json.dumps(event))
        f.write('\n')

# Create event file and initialize experiment
event_file = "./my_experiment_events.json"

# Write experiment start event
write_event(event_file, "experimentStart", {
    "task": "binary",
    "datasets": [{"kind": "Train", "shape": [1000, 20], "memory": 160000}],
    "steps": [
        {"index": 0, "name": "preprocessing", "type": "CustomStep", "status": "wait"},
        {"index": 1, "name": "model_search", "type": "SearchStep", "status": "wait"}
    ]
})

# Write step progress events
write_event(event_file, "stepStart", {"index": 0, "status": "process"})
write_event(event_file, "stepEnd", {"index": 0, "status": "finish"})

# Write trial results
write_event(event_file, "trialEnd", {
    "stepIndex": 1,
    "trialData": {
        "trialNo": 1,
        "maxTrials": 50,
        "reward": 0.95,
        "elapsed": 5.2,
        "hyperParams": {"n_estimators": "100", "max_depth": "10"}
    }
})
```

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

### NDJSON Event File Format Examples

Source: https://context7.com/datacanvasio/hyperboard/llms.txt

Illustrates the newline-delimited JSON (NDJSON) format for experiment events, including examples for experiment start, step progression, trial results, early stopping, and experiment completion.

```json
// Event file format: one JSON object per line

// experimentStart - Initialize experiment with configuration
{"type": "experimentStart", "payload": {"task": "regression", "datasets": [{"kind": "Train", "task": "regression", "shape": [404, 16], "memory": 51840}], "steps": [{"index": 0, "name": "data_clean", "type": "DataCleanStep", "status": "wait", "configuration": {"cv": true}}], "evaluation_metric": null}}

// stepStart - Mark step as running
{"type": "stepStart", "payload": {"index": 0, "status": "process", "start_datetime": 1645519525.487}}

// stepEnd - Mark step complete with results
{"type": "stepEnd", "payload": {"index": 0, "name": "data_clean", "type": "DataCleanStep", "status": "finish", "extension": {"features": {"inputs": ["col1", "col2"], "outputs": ["col1"], "reduced": ["col2"]}}, "end_datetime": 1645519525.602}}

// trialEnd - Record hyperparameter trial results
{"type": "trialEnd", "payload": {"stepIndex": 2, "trialData": {"trialNo": 1, "maxTrials": 10, "hyperParams": {"learning_rate": "0.1", "max_depth": "5"}, "models": [{"fold": 0, "importances": [{"name": "feature1", "imp": 0.65}]}], "reward": 3.21, "elapsed": 2.24, "is_cv": true, "metricName": "rmse"}}}

// earlyStopped - Early stopping triggered
{"type": "earlyStopped", "payload": {"stepIndex": 2, "data": {"reason": "max_no_improved_trials", "best_reward": 3.21}}}

// experimentEnd - Experiment complete
{"type": "experimentEnd", "payload": {}}
```

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

### Install hboard_widget package

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard-widget/js/README.md

Use this command to install the required widget package for Hyperboard in a Node.js environment.

```bash
npm install --save hboard_widget
```

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

### Build HyperBoard from source

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard/README.md

Commands to build the frontend assets and install the Python package.

```shell
cd hboard/js

# build frontend
yarn
yarn build
rm -rf ../hboard/assets/
cp -r build/ ../hboard/assets/

# install 
cd ..
python setup.py install
```

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

### Visualize experiment data

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard/README.md

Start the web server and load a specific experiment JSON file.

```shell
cd HyperBoard
hboard server --event-file=hboard/hboard/tests/events_example.json
```

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

### Create and Start WebApp

Source: https://context7.com/datacanvasio/hyperboard/llms.txt

Create a Tornado-based HTTP server to monitor experiment event files and serve a real-time dashboard. The server automatically handles port conflicts. For non-blocking execution, use WebAppRunner in a separate thread.

```python
from hboard.app import WebApp, WebAppRunner

# Create a web application to monitor an experiment event file
# Parameters:
#   event_file: path to the experiment events JSON file
#   server_port: HTTP server port (default: 8888)
#   port_retries: number of alternate ports to try if port is in use (default: 50)

webapp = WebApp(
    event_file="./events/experiment_events.json",
    server_port=8888,
    port_retries=50
)

# Start the server (blocking call)
webapp.start()
# Server logs: "experiment visualization http server is running at: http://0.0.0.0:8888"

# For non-blocking execution, use WebAppRunner in a separate thread
webapp = WebApp(event_file="./events/experiment_events.json", server_port=8888)
runner = WebAppRunner(webapp)
runner.start()  # Starts in background thread

# Access dashboard at http://localhost:8888

# Stop the server when done
runner.stop()
```

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

### Build hboard-widget from source

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard-widget/README.md

Install the hboard-widget in editable mode and enable Jupyter extensions after cloning the source code.

```bash
cd ./hboard-widget
pip install -e .
jupyter nbextension install --py --symlink --overwrite --sys-prefix hboard_widget
jupyter nbextension enable --py --sys-prefix hboard_widget
jupyter labextension develop --overwrite hboard_widget
```

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

### WebVisExperimentCallback Integration

Source: https://context7.com/datacanvasio/hyperboard/llms.txt

Integrate `WebVisExperimentCallback` with Hypernets experiments to automatically start a web server and stream events in real-time. It handles event file creation, web server lifecycle, and event streaming.

```python
from hboard.callbacks import WebVisExperimentCallback
from hypernets.experiment import make_experiment
from hypernets.examples.plain_model import PlainModel, PlainSearchSpace
from hypernets.tabular.datasets import dsutils
from sklearn.model_selection import train_test_split

# Load dataset
df = dsutils.load_boston()
df_train, df_eval = train_test_split(df, test_size=0.2)

# Create the visualization callback
# Parameters:
#   event_file_dir: directory to store event log files (default: "./events")
#   server_port: HTTP server port (default: 8888)

```

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

### Install hboard-widget with conda

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard-widget/README.md

Use this command to install the hboard-widget package using conda from the conda-forge channel.

```shell
conda install -c conda-forge hboard-widget
```

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

### Events API Endpoint

Source: https://context7.com/datacanvasio/hyperboard/llms.txt

Fetch experiment events from the web server using the REST API. The frontend polls this endpoint for real-time dashboard updates. You can fetch all events or events starting from a specific index.

```bash
# Fetch all events from the beginning
curl http://localhost:8888/api/events

# Response format:
# {"code": 0, "data": {"events": [
#   {"type": "experimentStart", "payload": {...}},
#   {"type": "stepStart", "payload": {...}},
#   {"type": "trialEnd", "payload": {...}},
#   ...
# ]}}

# Fetch events starting from a specific index (for incremental updates)
curl "http://localhost:8888/api/events?begin=5"

# Response contains only events from index 5 onwards
# {"code": 0, "data": {"events": [{"type": "stepEnd", ...}, ...]}}
```

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

### Configure and Create Experiment

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard-widget/hboard_widget/examples/01.visual_experiment.ipynb

Loads the dataset, splits it into training and evaluation sets, and initializes the experiment with a specific search space.

```python
df = dsutils.load_boston()
df_train, df_eval = train_test_split(df, test_size=0.2)
search_space = PlainSearchSpace(enable_lr=False, enable_nn=False, enable_dt=False, enable_dtr=True)
experiment = make_experiment(PlainModel, df_train,
                             target='target',
                             search_space=search_space,
                             callbacks=[],
                             search_callbacks=[])
```

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

### Initialize Hypernets Experiment Environment

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard-widget/hboard_widget/examples/01.visual_experiment.ipynb

Imports necessary libraries and modules for setting up a Hypernets experiment.

```python
import warnings
warnings.filterwarnings('ignore')

from sklearn.model_selection import train_test_split
from hypernets.examples.plain_model import PlainModel, PlainSearchSpace
from hypernets.experiment import make_experiment
from hypernets.tabular.datasets import dsutils
```

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

### Create event file

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard/README.md

Initialize an empty file to store experiment state events.

```shell
touch events.txt
```

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

### Display Experiment Process Widget

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard-widget/hboard_widget/examples/01.visual_experiment.ipynb

Runs the experiment and displays the ExperimentProcessWidget to monitor the training progress.

```python
from hboard_widget import ExperimentProcessWidget
estimator = experiment.run(max_trials=3)

widget = ExperimentProcessWidget(experiment)
display(widget)
```

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

### Display Experiment Summary Widget

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard-widget/hboard_widget/examples/01.visual_experiment.ipynb

Initializes and displays the ExperimentSummary widget for an existing experiment object.

```python
from hboard_widget import ExperimentSummary
experiment_summary_widget = ExperimentSummary(experiment)
display(experiment_summary_widget)
```

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

### Configure WebVisExperimentCallback for Experiment Monitoring

Source: https://context7.com/datacanvasio/hyperboard/llms.txt

Use WebVisExperimentCallback to enable a web-based dashboard for experiment tracking. The dashboard is accessible at the specified server port during execution.

```python
web_callback = WebVisExperimentCallback(
    event_file_dir="./experiment_logs",
    server_port=8888,
    exit_web_server_on_finish=False
)

# Create search space
search_space = PlainSearchSpace(
    enable_lr=True,
    enable_nn=False,
    enable_dt=True,
    enable_dtr=True
)

# Create experiment with visualization callback
experiment = make_experiment(
    PlainModel,
    df_train,
    target='target',
    search_space=search_space,
    callbacks=[web_callback],  # Add visualization callback
    search_callbacks=[]
)

# Run experiment - dashboard automatically available at http://localhost:8888
estimator = experiment.run(max_trials=20)
```

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

### Display hboard help

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard/README.md

View command-line usage information for the hboard tool.

```shell
hboard -h
```

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

### Display Dataset Summary Widget

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard-widget/hboard_widget/examples/01.visual_experiment.ipynb

Initializes and displays the DatasetSummary widget to visualize dataset information.

```python
from hboard_widget import DatasetSummary
dataset_summary_widget = DatasetSummary(experiment)
display(dataset_summary_widget)
```

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

### Display Experiment Configuration with ExperimentSummary

Source: https://context7.com/datacanvasio/hyperboard/llms.txt

The ExperimentSummary widget provides an inline view of experiment metadata and configuration within a Jupyter notebook.

```python
import warnings
warnings.filterwarnings('ignore')

from sklearn.model_selection import train_test_split
from hypernets.examples.plain_model import PlainModel, PlainSearchSpace
from hypernets.experiment import make_experiment
from hypernets.tabular.datasets import dsutils
from hboard_widget import ExperimentSummary

# Load and prepare data
df = dsutils.load_boston()
df_train, df_eval = train_test_split(df, test_size=0.2)

# Create search space
search_space = PlainSearchSpace(
    enable_lr=False,
    enable_nn=False,
    enable_dt=False,
    enable_dtr=True
)

# Create experiment
experiment = make_experiment(
    PlainModel,
    df_train,
    target='target',
    search_space=search_space,
    callbacks=[],
    search_callbacks=[]
)

# Display experiment configuration widget
# Shows: task type, dataset shape, memory usage, pipeline steps, early stopping config
experiment_summary_widget = ExperimentSummary(experiment)
display(experiment_summary_widget)
```

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

### Visualize dataset information with DatasetSummary widget

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard-widget/README.md

Instantiate and display the DatasetSummary widget to visualize dataset characteristics derived from the experiment's data.

```python
from hboard_widget import DatasetSummary
dataset_summary_widget = DatasetSummary(experiment.get_data_character())
display(dataset_summary_widget)
```

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

### Monitor Real-time Progress with ExperimentProcessWidget

Source: https://context7.com/datacanvasio/hyperboard/llms.txt

The ExperimentProcessWidget visualizes live experiment execution, including trial results and performance metrics.

```python
import warnings
warnings.filterwarnings('ignore')

from sklearn.model_selection import train_test_split
from hypernets.examples.plain_model import PlainModel, PlainSearchSpace
from hypernets.experiment import make_experiment
from hypernets.tabular.datasets import dsutils
from hboard_widget import ExperimentProcessWidget

# Load and prepare data
df = dsutils.load_boston()
df_train, df_eval = train_test_split(df, test_size=0.2)

# Create search space with multiple model types
search_space = PlainSearchSpace(
    enable_lr=True,
    enable_nn=False,
    enable_dt=True,
    enable_dtr=True
)

# Create experiment
experiment = make_experiment(
    PlainModel,
    df_train,
    target='target',
    search_space=search_space,
    callbacks=[],
    search_callbacks=[]
)

# Run experiment
estimator = experiment.run(max_trials=10)

# Display experiment process widget after completion
```

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

### Automatic Jupyter Visualization with NotebookExperimentCallback

Source: https://context7.com/datacanvasio/hyperboard/llms.txt

Integrates `NotebookExperimentCallback` into an experiment to automatically create and update visualization widgets in Jupyter notebooks during execution. Loads data, defines a search space, and runs the experiment.

```python
import warnings
warnings.filterwarnings('ignore')

from sklearn.model_selection import train_test_split
from hypernets.examples.plain_model import PlainModel, PlainSearchSpace
from hypernets.experiment import make_experiment
from hypernets.tabular.datasets import dsutils
from hboard_widget.callbacks import NotebookExperimentCallback

# Load and prepare data
df = dsutils.load_boston()
df_train, df_eval = train_test_split(df, test_size=0.2)

# Create the notebook visualization callback
notebook_callback = NotebookExperimentCallback()

# Create search space
search_space = PlainSearchSpace(
    enable_lr=True,
    enable_dt=True,
    enable_dtr=True
)

# Create experiment with notebook callback
experiment = make_experiment(
    PlainModel,
    df_train,
    target='target',
    search_space=search_space,
    callbacks=[notebook_callback],  # Add notebook visualization callback
    search_callbacks=[]
)

# Run experiment - widget automatically created and updated in real-time
estimator = experiment.run(max_trials=10)
```

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

### Clone Project Repository

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard-frontend/README.md

Clones the HyperBoard project repository from GitHub. This command should be run in your terminal.

```bash
git clone https://github.com/DataCanvasIO/HyperBoard.git
```

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

### Build JS for development

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard-widget/README.md

Rebuild the JavaScript frontend when making code changes in the js directory.

```bash
cd ./hboard-widget/js
yarn run build
```

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

### Display Experiment Widget

Source: https://context7.com/datacanvasio/hyperboard/llms.txt

Instantiates and displays an experiment widget for visualizing experiment progress and results.

```python
widget = ExperimentProcessWidget(experiment)
display(widget)
```

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

### Append experiment event

Source: https://github.com/datacanvasio/hyperboard/blob/main/hboard/README.md

Add a JSON-formatted experiment event to the event file.

```shell
echo '{"type": "experimentStart", "payload": {"task": "binary", "datasets": [{"kind": "Train", "task": "binary", "shape": [904, 17], "memory": 123072}], "steps": [{"index": 0, "name": "data_clean", "type": "DataCleanStep", "status": "wait", "configuration": {"cv": true, "data_cleaner_args": {"nan_chars": null, "correct_object_dtype": true, "drop_constant_columns": true, "drop_label_nan_rows": true, "drop_idness_columns": true, "drop_columns": null, "reserve_columns": null, "drop_duplicated_columns": false, "reduce_mem_usage": false, "int_convert_to": "float"}, "name": "data_clean", "train_test_split_strategy": null}, "extension": {}, "start_datetime": null, "end_datetime": null}, {"index": 1, "name": "space_searching", "type": "SpaceSearchStep", "status": "wait", "configuration": {"cv": true, "name": "space_searching", "num_folds": 3, "earlyStopping": {"enable": true, "exceptedReward": null, "maxNoImprovedTrials": 10, "timeLimit": 3600, "mode": "max"}}, "extension": {}, "start_datetime": null, "end_datetime": null}, {"index": 2, "name": "final_ensemble", "type": "EnsembleStep", "status": "wait", "configuration": {"ensemble_size": 20, "name": "final_ensemble", "scorer": "make_scorer(accuracy_score)"}, "extension": {}, "start_datetime": null, "end_datetime": null}], "evaluation_metric": null, "confusion_matrix": null, "resource_usage": null, "prediction_stats": null}}' >> events.txt
```

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