Differences

This shows you the differences between two versions of the page.

--- products:ict:python:machine_learning:light_gbm [2023/10/12 17:28] – created wikiadmin
+++ products:ict:python:machine_learning:light_gbm [2023/10/12 17:30] (current) – wikiadmin
@@ Line 4: / Line 4: @@
 . **Installation**:
-   You can install LightGBM using pip:
+You can install LightGBM using pip:
+pip install lightgbm
-   ```bash
-   pip install lightgbm
-   ```
 . **Importing LightGBM**:
-   Once installed, you can import LightGBM in your Python script or Jupyter Notebook:
+Once installed, you can import LightGBM in your Python script or Jupyter Notebook:
+import lightgbm as lgb
-   ```python
-   import lightgbm as lgb
-   ```
 . **Data Preparation**:
-   Before you can use LightGBM, you need to prepare your data. LightGBM works with tabular data, and it expects the data to be in a format that is compatible with the `Dataset` object provided by the library.
+Before you can use LightGBM, you need to prepare your data. LightGBM works with tabular data, and it expects the data to be in a format that is compatible with the `Dataset` object provided by the library.
    - Load and preprocess your dataset using libraries like Pandas and NumPy.
@@ Line 27: / Line 26: @@
 . **Creating a Dataset**:
-   To efficiently use LightGBM, you need to create a `Dataset` object from your data. This object is optimized for training and prediction.
+To efficiently use LightGBM, you need to create a `Dataset` object from your data. This object is optimized for training and prediction.
+train_data = lgb.Dataset(data=X_train, label=y_train)
+test_data = lgb.Dataset(data=X_test, label=y_test, reference=train_data)
-   ```python
-   train_data = lgb.Dataset(data=X_train, label=y_train)
-   test_data = lgb.Dataset(data=X_test, label=y_test, reference=train_data)
-   ```
 . **Setting Parameters**:
-   LightGBM has a wide range of parameters that control the training process and the model's behavior. Some important parameters include:
+LightGBM has a wide range of parameters that control the training process and the model's behavior. Some important parameters include:
-   - `objective`: Specifies the learning task (e.g., "regression," "binary," or "multiclass").
+- `objective`: Specifies the learning task (e.g., "regression," "binary," or "multiclass").
-   - `num_leaves`: Number of leaves in each tree. It controls the complexity of the model.
-   - `learning_rate`: Step size for updates during training.
-   - `max_depth`: Maximum depth of the tree.
-   - `num_boost_round`: Number of boosting iterations (trees).
-   - `metric`: Evaluation metric for model performance.
-   You can set these parameters in a dictionary and pass it to the training process.
+- `num_leaves`: Number of leaves in each tree. It controls the complexity of the model.
+- `learning_rate`: Step size for updates during training.
+- `max_depth`: Maximum depth of the tree.
+- `num_boost_round`: Number of boosting iterations (trees).
+- `metric`: Evaluation metric for model performance.
+You can set these parameters in a dictionary and pass it to the training process.
 . **Training the Model**:
-   To train the LightGBM model, you use the `train` method, passing in your training data and the parameter dictionary:
+To train the LightGBM model, you use the `train` method, passing in your training data and the parameter dictionary:
-   ```python
-   num_round = 100
-   bst = lgb.train(params, train_data, num_round, valid_sets=[test_data], early_stopping_rounds=10)
-   ```
-   The `early_stopping_rounds` parameter allows the training to stop early if the evaluation metric doesn't improve for a specified number of rounds on the validation set.
+num_round = 100
+bst = lgb.train(params, train_data, num_round, valid_sets=[test_data], early_stopping_rounds=10)
+The `early_stopping_rounds` parameter allows the training to stop early if the evaluation metric doesn't improve for a specified number of rounds on the validation set.
 . **Making Predictions**:
-   After training, you can use the trained model to make predictions on new data:
+After training, you can use the trained model to make predictions on new data:
+predictions = bst.predict(X_new_data)
-   ```python
-   predictions = bst.predict(X_new_data)
-   ```
 . **Model Evaluation**:
-   Evaluate the model's performance using various metrics. You can access the model's feature importance and even plot the trees in the model to gain insights into its decision-making process.
+Evaluate the model's performance using various metrics. You can access the model's feature importance and even plot the trees in the model to gain insights into its decision-making process.
 . **Hyperparameter Tuning**:
-   It's common to perform hyperparameter tuning to find the best set of parameters for your specific problem. Techniques like grid search or random search can be used.
+It's common to perform hyperparameter tuning to find the best set of parameters for your specific problem. Techniques like grid search or random search can be used.
 . **Deployment**:
-    Once you're satisfied with the model's performance, you can deploy it for real-world predictions, either in a production environment or in your applications.
+Once you're satisfied with the model's performance, you can deploy it for real-world predictions, either in a production environment or in your applications.
 LightGBM's efficiency and speed make it an excellent choice for large datasets and computationally intensive tasks. However, it's essential to understand its parameters and the data you are working with to achieve optimal results.