PySurvival

https://github.com/square/pysurvival
Admin
PySurvival is an open source Python package for survival analysis modeling that helps predict when...
Tokens:78,041
Snippets:744
Trust Score:9.4
License:Apache-2.0
Update:6 months ago
Show doc for...
Tokens:
Raw
### Example: Initialize SimulationModel

Source: https://github.com/square/pysurvival/blob/master/docs/models/simulations.html

Basic setup for initializing a simulation model with a Gompertz distribution.

```python
import pandas as pd
from pysurvival.models.simulations import SimulationModel
%pylab inline

# Initializing the simulation model
sim = SimulationModel( survival_distribution = 'gompertz',
                       risk_type = 'linear',
```

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

### SmoothKaplanMeierModel Example Usage

Source: https://github.com/square/pysurvival/blob/master/docs/models/smooth_kaplan_meier.html

Example demonstrating how to initialize, fit, and display the SmoothKaplanMeierModel.

```APIDOC
## Example Usage

### Description
This example demonstrates how to initialize the `SmoothKaplanMeierModel`, fit it with sample data, and visualize the survival function and its confidence intervals.

### Code
```python
# Importing modules
import numpy as np
from matplotlib import pyplot as plt
from pysurvival.models.non_parametric import SmoothKaplanMeierModel
from pysurvival.utils.display import display_non_parametric
# %matplotlib inline # Uncomment when using Jupyter

# Generating random times and event indicators
T = np.round(np.abs(np.random.normal(10, 10, 1000)), 1)
E = np.random.binomial(1, 0.3, 1000)

# Initializing the SmoothKaplanMeierModel
skm_model = SmoothKaplanMeierModel(bandwidth=1., kernel='Cosine')

# Fitting the model and display the survival function and confidence intervals
skm_model.fit(T, E, alpha=0.95)

# Displaying the survival function and confidence intervals
display_non_parametric(skm_model)

# Figure 1 - Representation of the Smooth Kaplan Meier Survival function
# ![PySurvival - Smooth Kaplan Meier - Representing the Survival function](images/smooth_kaplan_meier.png "PySurvival - Smooth Kaplan Meier - Representing the Survival function")
```
```

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

### Importing required packages

Source: https://github.com/square/pysurvival/blob/master/docs/models/nonlinear_coxph.html

Initial setup for data manipulation, visualization, and model training.

```python
import numpy as np
import pandas as pd
from matplotlib import pyplot as plt
from sklearn.model_selection import train_test_split
from pysurvival.models.simulations import SimulationModel
from pysurvival.models.semi_parametric import NonLinearCoxPHModel
from pysurvival.utils.metrics import concordance_index
from pysurvival.utils.display import integrated_brier_score
#%pylab inline
```

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

### Install PySurvival using pip

Source: https://github.com/square/pysurvival/blob/master/README.md

Use this command to install PySurvival. Ensure you have a working GCC installation.

```bash
pip install pysurvival
```

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

### Importing PySurvival and dependencies

Source: https://github.com/square/pysurvival/blob/master/docs/models/conditional_survival_forest.html

Initial setup for the modeling environment.

```python
import numpy as np
import pandas as pd
from matplotlib import pyplot as plt
from sklearn.model_selection import train_test_split
from pysurvival.models.simulations import SimulationModel
from pysurvival.models.survival_forest import ConditionalSurvivalForestModel
from pysurvival.utils.metrics import concordance_index
from pysurvival.utils.display import integrated_brier_score
%pylab inline
```

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

### Install GCC on Ubuntu

Source: https://github.com/square/pysurvival/blob/master/docs/installation.html

Install GCC version 8 and its C++ counterpart on Ubuntu. Adjust the version number if a newer GCC is available.

```bash
sudo apt install gcc-8 g++-8
```

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

### Importing PySurvival and dependencies

Source: https://github.com/square/pysurvival/blob/master/docs/models/random_survival_forest.html

Initial setup for the modeling environment, including data manipulation, plotting, and survival analysis utilities.

```python
import numpy as np
import pandas as pd
from matplotlib import pyplot as plt
from sklearn.model_selection import train_test_split
from pysurvival.models.simulations import SimulationModel
from pysurvival.models.survival_forest import RandomSurvivalForestModel
from pysurvival.utils.metrics import concordance_index
from pysurvival.utils.display import integrated_brier_score
%pylab inline
```

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

### PySurvival Quick Modeling Example

Source: https://github.com/square/pysurvival/blob/master/README.md

This example demonstrates building and evaluating both CoxPH and MTLR models using PySurvival. It includes loading data, fitting models, and calculating concordance index.

```python
# Loading the modules
from pysurvival.models.semi_parametric import CoxPHModel
from pysurvival.models.multi_task import LinearMultiTaskModel
from pysurvival.datasets import Dataset
from pysurvival.utils.metrics import concordance_index

# Loading and splitting a simple example into train/test sets
X_train, T_train, E_train, X_test, T_test, E_test = \
	Dataset('simple_example').load_train_test()

# Building a CoxPH model
coxph_model = CoxPHModel()
coxph_model.fit(X=X_train, T=T_train, E=E_train, init_method='he_uniform', 
                l2_reg = 1e-4, lr = .4, tol = 1e-4)

# Building a MTLR model
mtlr = LinearMultiTaskModel()
mtlr.fit(X=X_train, T=T_train, E=E_train, init_method = 'glorot_uniform', 
           optimizer ='adam', lr = 8e-4)

# Checking the model performance
c_index1 = concordance_index(model=coxph_model, X=X_test, T=T_test, E=E_test )
print("CoxPH model c-index = {:.2f}".format(c_index1))

c_index2 = concordance_index(model=mtlr, X=X_test, T=T_test, E=E_test )
print("MTLR model c-index = {:.2f}".format(c_index2))
```

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

### Compare actual and predicted values using compare_to_actual

Source: https://github.com/square/pysurvival/blob/master/docs/miscellaneous/tips.html

Examples showing how to compare predicted and actual values for loans that were fully repaid versus those that are still active.

```python
from pysurvival.utils.display import compare_to_actual
results = compare_to_actual(neural_mtlr, X_test, T_test, E_test,
                            is_at_risk = False,  figure_size=(16, 6),
                            metrics = ['rmse', 'mean', 'median'])
```

```python
results = compare_to_actual(neural_mtlr, X_test, T_test, E_test,
                            is_at_risk = True,  figure_size=(16, 6),
                            metrics = ['rmse', 'mean', 'median'])
```

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

### Example Usage of SmoothKaplanMeierModel

Source: https://github.com/square/pysurvival/blob/master/docs/models/smooth_kaplan_meier.html

Demonstrates the complete workflow of using the SmoothKaplanMeierModel, including data generation, model initialization, fitting, and displaying the survival function with confidence intervals.

```python
# Importing modules
import numpy as np
from matplotlib import pyplot as plt
from pysurvival.models.non_parametric import SmoothKaplanMeierModel
from pysurvival.utils.display import display_non_parametric
# %matplotlib inline #Uncomment when using Jupyter 

# Generating random times and event indicators 
T = np.round(np.abs(np.random.normal(10, 10, 1000)), 1)
E = np.random.binomial(1, 0.3, 1000)

# Initializing the SmoothKaplanMeierModel
skm_model = SmoothKaplanMeierModel(bandwidth=1., kernel='Cosine')

# Fitting the model and display the survival function and confidence intervals
skm_model.fit(T, E, alpha=0.95)

# Displaying the survival function and confidence intervals
display_non_parametric(skm_model)
```

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

### Install GCC on MacOS

Source: https://github.com/square/pysurvival/blob/master/docs/installation.html

Use this command to install the latest version of GCC on MacOS if you have Homebrew installed.

```bash
brew install gcc
```

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

### List GCC Files on MacOS

Source: https://github.com/square/pysurvival/blob/master/docs/installation.html

This command lists all installed GCC files, which is useful for identifying the correct paths for environment variables.

```bash
brew ls gcc
```

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

### Install GCC on CentOS7

Source: https://github.com/square/pysurvival/blob/master/docs/installation.html

Install GCC version 8 and its C++ counterpart on CentOS 7 using yum. Adjust the version number if a newer GCC is available.

```bash
sudo yum install centos-release-scl
sudo yum install devtoolset-8-gcc devtoolset-8-gcc-c++  -y
```

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

### Set C++ Compiler Environment Variables on Ubuntu

Source: https://github.com/square/pysurvival/blob/master/docs/installation.html

Set the C++ compiler and C compiler paths for Ubuntu. Ensure the version number matches your GCC installation.

```bash
export CXX=/usr/bin/g++-8
export CC=/usr/bin/gcc-8
```

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

### Get Model Parameters

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Retrieves the estimated parameters of a fitted survival model. These parameters define the shape and scale of the survival distribution.

```python
from pysurvival.models.parametric import Weibull
from pysurvival.utils.data import simulate_data

# Assume weibull_model is already fitted with time, event, X
# time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
# weibull_model.fit(time, event, X)

# Get the model parameters
# params = weibull_model.get_parameters()
# print(params)

```

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

### Get Model Parameters for Survival Forest

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Retrieves the parameters of a fitted Survival Forest model. This typically includes information about the trees, such as their structure and split points.

```python
from pysurvival.models.non_parametric import SurvivalForest
from pysurvival.utils.data import simulate_data

# Simulate survival data
time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)

# Initialize and fit a Survival Forest model
survival_forest = SurvivalForest()
survival_forest.fit(time, event, X)

# Get the model parameters
# params = survival_forest.get_params()
# print(params)

```

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

### Get Model Summary for CoxPH

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Provides a summary of the fitted CoxPH model, including coefficients, standard errors, p-values, and overall model fit statistics.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import simulate_data

# Assume coxph_model is already fitted with time, event, X
# time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
# coxph_model.fit(time, event, X)

# Get the model summary
# summary = coxph_model.summary()
# print(summary)

```

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

### Get Model Parameters for CoxPH

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Retrieves the estimated coefficients (log hazard ratios) from a fitted CoxPH model. These coefficients quantify the effect of each covariate on the hazard rate.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import simulate_data

# Assume coxph_model is already fitted with time, event, X
# time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
# coxph_model.fit(time, event, X)

# Get the model parameters (coefficients)
# params = coxph_model.get_params()
# print(params)

```

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

### Get Model Summary for Survival Forest

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Provides a summary of the fitted Survival Forest model, which may include information about the number of trees, feature importances, and potentially performance metrics.

```python
from pysurvival.models.non_parametric import SurvivalForest
from pysurvival.utils.data import simulate_data

# Simulate survival data
time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)

# Initialize and fit a Survival Forest model
survival_forest = SurvivalForest()
survival_forest.fit(time, event, X)

# Get the model summary
# summary = survival_forest.summary()
# print(summary)

```

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

### Get Baseline Survival Function from CoxPH

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Retrieves the estimated baseline survival function from a fitted CoxPH model. This represents the survival probability when all covariates are zero.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import simulate_data

# Assume coxph_model is already fitted with time, event, X
# time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
# coxph_model.fit(time, event, X)

# Get the baseline survival function (times and values)
# baseline_survival_times, baseline_survival_values = coxph_model.baseline_survival_function
# print(baseline_survival_times)
# print(baseline_survival_values)

```

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

### Get Survival Times from Survival Forest

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Retrieves the predicted survival times from a fitted Survival Forest model. This provides an estimate of the time until the event occurs for each individual.

```python
from pysurvival.models.non_parametric import SurvivalForest
from pysurvival.utils.data import simulate_data

# Simulate survival data
time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)

# Initialize and fit a Survival Forest model
survival_forest = SurvivalForest()
survival_forest.fit(time, event, X)

# Predict survival times for new data (e.g., X_new)
# survival_times = survival_forest.predict_time(X_new)

```

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

### Get Model Parameters for DeepSurv

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Retrieves the learned parameters (weights and biases) of a fitted DeepSurv model. These parameters define the neural network's structure and learned relationships.

```python
from pysurvival.models.deep_surv import DeepSurv
from pysurvival.utils.data import simulate_data

# Simulate survival data
time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)

# Initialize and fit a DeepSurv model
deepsurv_model = DeepSurv()
deepsurv_model.fit(time, event, X)

# Get the model parameters (weights and biases)
# params = deepsurv_model.get_params()
# print(params)

```

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

### Get Model Summary for DeepSurv

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Provides a summary of the fitted DeepSurv model, typically including information about the training process, loss values, and potentially performance metrics like the C-index.

```python
from pysurvival.models.deep_surv import DeepSurv
from pysurvival.utils.data import simulate_data

# Simulate survival data
time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)

# Initialize and fit a DeepSurv model
deepsurv_model = DeepSurv()
deepsurv_model.fit(time, event, X)

# Get the model summary
# summary = deepsurv_model.summary()
# print(summary)

```

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

### Get Baseline Hazard Function from CoxPH

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Retrieves the estimated baseline hazard function from a fitted CoxPH model. This represents the hazard rate when all covariates are zero.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import simulate_data

# Assume coxph_model is already fitted with time, event, X
# time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
# coxph_model.fit(time, event, X)

# Get the baseline hazard function (times and values)
# baseline_hazard_times, baseline_hazard_values = coxph_model.baseline_hazard_function
# print(baseline_hazard_times)
# print(baseline_hazard_values)

```

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

### Get Survival Function for CoxPH with Baseline

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Calculates the survival function for a given set of covariates using the baseline survival function from a fitted CoxPH model. This accounts for the effect of covariates.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import simulate_data
import numpy as np

# Assume coxph_model is already fitted with time, event, X
# time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
# coxph_model.fit(time, event, X)

# Define a specific covariate vector
# X_specific = X[0, :]

# Get the survival function at specific time points
# time_points = np.array([10, 50, 100])
# survival_at_times = coxph_model.get_survival_function(X_specific, time_points)
# print(survival_at_times)

```

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

### Get Model Summary for Parametric Models

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Provides a summary of the fitted parametric survival model, including estimated parameters, standard errors, and convergence information. This helps in assessing model fit and parameter significance.

```python
from pysurvival.models.parametric import Weibull
from pysurvival.utils.data import simulate_data

# Assume weibull_model is already fitted with time, event, X
# time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
# weibull_model.fit(time, event, X)

# Get the model summary
# summary = weibull_model.summary()
# print(summary)

```

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

### Get Hazard Ratios from CoxPH Model

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Retrieves the hazard ratios from a fitted CoxPH model. Hazard ratios indicate the multiplicative change in the hazard rate for a one-unit increase in a covariate.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import simulate_data

# Assume coxph_model is already fitted with time, event, X
# time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
# coxph_model.fit(time, event, X)

# Get hazard ratios
# hazard_ratios = coxph_model.hazard_ratios
# print(hazard_ratios)

```

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

### Get Hazard Function for CoxPH with Baseline

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Calculates the hazard function for a given set of covariates using the baseline hazard function from a fitted CoxPH model. This accounts for the effect of covariates on the instantaneous risk.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import simulate_data
import numpy as np

# Assume coxph_model is already fitted with time, event, X
# time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
# coxph_model.fit(time, event, X)

# Define a specific covariate vector
# X_specific = X[0, :]

# Get the hazard function at specific time points
# time_points = np.array([10, 50, 100])
# hazard_at_times = coxph_model.get_hazard_function(X_specific, time_points)
# print(hazard_at_times)

```

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

### Initialize Application

Source: https://github.com/square/pysurvival/blob/master/docs/models/conditional_survival_forest.html

Initializes the application with version and base URL information. This is a common pattern for web applications.

```javascript
app.initialize({version:"1.0.4",url:{base:".."}})
```

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

### Get Survival Function for Multiple Individuals with CoxPH and SurvivalDataFrame

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Calculates survival functions for multiple individuals using a fitted CoxPH model and a SurvivalDataFrame containing their covariate information. This allows for efficient comparison of survival trajectories.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import SurvivalDataFrame
from pysurvival.utils.data import simulate_data

# Simulate survival data and create a SurvivalDataFrame
time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
survival_df = SurvivalDataFrame(X, time, event)

# Assume coxph_model is already fitted with survival_df
# coxph_model.fit(survival_df)

# Create a SurvivalDataFrame for prediction (features only)
# X_new_multi, _, _ = simulate_data(n_samples=5, n_features=5, n_informative=3)
# survival_df_new_multi = SurvivalDataFrame(X_new_multi)

# Predict survival functions for multiple individuals
# survival_probabilities_multi = coxph_model.predict_survival_function(survival_df_new_multi)

```

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

### Get Baseline Hazard Function from CoxPH with SurvivalDataFrame

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Retrieves the estimated baseline hazard function from a fitted CoxPH model that was trained using a SurvivalDataFrame. This represents the hazard rate when all covariates are zero.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import SurvivalDataFrame
from pysurvival.utils.data import simulate_data

# Simulate survival data and create a SurvivalDataFrame
time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
survival_df = SurvivalDataFrame(X, time, event)

# Assume coxph_model is already fitted with survival_df
# coxph_model.fit(survival_df)

# Get the baseline hazard function (times and values)
# baseline_hazard_times, baseline_hazard_values = coxph_model.baseline_hazard_function
# print(baseline_hazard_times)
# print(baseline_hazard_values)

```

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

### Get Survival Function for CoxPH with Baseline and SurvivalDataFrame

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Calculates the survival function for a given covariate vector (from a SurvivalDataFrame) using the baseline survival function from a fitted CoxPH model. This accounts for the effect of covariates.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import SurvivalDataFrame
from pysurvival.utils.data import simulate_data
import numpy as np

# Simulate survival data and create a SurvivalDataFrame
time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
survival_df = SurvivalDataFrame(X, time, event)

# Assume coxph_model is already fitted with survival_df
# coxph_model.fit(survival_df)

# Create a SurvivalDataFrame for prediction (features only)
# X_specific_df = SurvivalDataFrame(X[0, :].reshape(1, -1))

# Define specific time points for prediction
# time_points = np.array([10, 50, 100])

# Get the survival function at specific time points
# survival_at_times = coxph_model.get_survival_function(X_specific_df, time_points)
# print(survival_at_times)

```

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

### Get Hazard Function for Multiple Individuals with CoxPH and SurvivalDataFrame

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Calculates hazard functions for multiple individuals using a fitted CoxPH model and a SurvivalDataFrame containing their covariate information. This enables efficient comparison of instantaneous risk profiles.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import SurvivalDataFrame
from pysurvival.utils.data import simulate_data

# Simulate survival data and create a SurvivalDataFrame
time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
survival_df = SurvivalDataFrame(X, time, event)

# Assume coxph_model is already fitted with survival_df
# coxph_model.fit(survival_df)

# Create a SurvivalDataFrame for prediction (features only)
# X_new_multi, _, _ = simulate_data(n_samples=5, n_features=5, n_informative=3)
# survival_df_new_multi = SurvivalDataFrame(X_new_multi)

# Predict hazard functions for multiple individuals
# hazard_rates_multi = coxph_model.predict_hazard(survival_df_new_multi)

```

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

### Get Baseline Survival Function from CoxPH with SurvivalDataFrame

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Retrieves the estimated baseline survival function from a fitted CoxPH model that was trained using a SurvivalDataFrame. This represents the survival probability when all covariates are zero.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import SurvivalDataFrame
from pysurvival.utils.data import simulate_data

# Simulate survival data and create a SurvivalDataFrame
time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
survival_df = SurvivalDataFrame(X, time, event)

# Assume coxph_model is already fitted with survival_df
# coxph_model.fit(survival_df)

# Get the baseline survival function (times and values)
# baseline_survival_times, baseline_survival_values = coxph_model.baseline_survival_function
# print(baseline_survival_times)
# print(baseline_survival_values)

```

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

### Initialize and fit KaplanMeierModel

Source: https://github.com/square/pysurvival/blob/master/docs/models/kaplan_meier.html

Demonstrates importing the model, generating synthetic data, fitting the model, and displaying the survival function.

```python
# Importing modules
import numpy as np
from matplotlib import pyplot as plt
from pysurvival.models.non_parametric import KaplanMeierModel
from pysurvival.utils.display import display_non_parametric
# %matplotlib inline #Uncomment when using Jupyter 

# Generating random times and event indicators 
T = np.round(np.abs(np.random.normal(10, 10, 1000)), 1)
E = np.random.binomial(1, 0.3, 1000)

# Initializing the KaplanMeierModel
km_model = KaplanMeierModel()

# Fitting the model 
km_model.fit(T, E, alpha=0.95)

# Displaying the survival function and confidence intervals
display_non_parametric(km_model)
```

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

### Get Hazard Function for CoxPH with Baseline and SurvivalDataFrame

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Calculates the hazard function for a given covariate vector (from a SurvivalDataFrame) using the baseline hazard function from a fitted CoxPH model. This accounts for the effect of covariates on the instantaneous risk.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import SurvivalDataFrame
from pysurvival.utils.data import simulate_data
import numpy as np

# Simulate survival data and create a SurvivalDataFrame
time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
survival_df = SurvivalDataFrame(X, time, event)

# Assume coxph_model is already fitted with survival_df
# coxph_model.fit(survival_df)

# Create a SurvivalDataFrame for prediction (features only)
# X_specific_df = SurvivalDataFrame(X[0, :].reshape(1, -1))

# Define specific time points for prediction
# time_points = np.array([10, 50, 100])

# Get the hazard function at specific time points
# hazard_at_times = coxph_model.get_hazard_function(X_specific_df, time_points)
# print(hazard_at_times)

```

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

### LinearSVMModel Initialization and Fitting

Source: https://github.com/square/pysurvival/blob/master/docs/models/linear_svm.html

This snippet shows how to initialize and fit the LinearSVMModel with various parameters.

```APIDOC
## LinearSVMModel Initialization and Fitting

### Description
Initializes and fits the LinearSVMModel to the provided survival data.

### Method
`fit(X, T, E, init_method='glorot_uniform', lr=1e-4, max_iter=100, l2_reg=1e-4, alpha=0.95, tol=1e-3, verbose=True)`

### Parameters
#### Input Data
- **X** (array-like) - Input samples (n_samples, n_features).
- **T** (array-like) - Target values (event or censoring times).
- **E** (array-like) - Event indicators (1 for event, 0 for censoring).

#### Model Parameters
- **init_method** (str, default='glorot_uniform') - Initialization method for weights. Options include 'glorot_uniform', 'he_uniform', 'uniform', 'glorot_normal', 'he_normal', 'normal', 'ones', 'zeros', 'orthogonal'.
- **lr** (float, default=1e-4) - Learning rate for optimization.
- **max_iter** (int, default=100) - Maximum number of iterations for the optimization.
- **l2_reg** (float, default=1e-4) - L2 regularization parameter.
- **alpha** (float, default=0.95) - Confidence level.
- **tol** (float, default=1e-3) - Tolerance for stopping criteria.
- **verbose** (bool, default=True) - Whether to output detailed logging.

#### Optional Parameters
- **with_bias** (bool, default=True) - Whether to include a bias term.

### Returns
- self : object - The fitted LinearSVMModel instance.

### Request Example
```python
from pysurvival.models.svm import LinearSVMModel

svm_model = LinearSVMModel()
svm_model.fit(X_train, T_train, E_train, init_method='he_uniform', lr=1e-3, max_iter=200)
```
```

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

### Initialize SmoothKaplanMeierModel

Source: https://github.com/square/pysurvival/blob/master/docs/models/smooth_kaplan_meier.html

How to instantiate the Smooth Kaplan Meier model class.

```APIDOC
## Instance

### Description
To create an instance of the Smooth Kaplan Meier model, use the `pysurvival.models.non_parametric.SmoothKaplanMeierModel` class.

### Usage
```python
from pysurvival.models.non_parametric import SmoothKaplanMeierModel
model = SmoothKaplanMeierModel()
```
```

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

### Get Feature Importance for DeepSurv

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Retrieves feature importance scores from a fitted DeepSurv model. This requires specific methods or analysis of the trained neural network, as it's not a direct output like in tree-based models.

```python
from pysurvival.models.deep_surv import DeepSurv
from pysurvival.utils.data import simulate_data

# Simulate survival data
time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)

# Initialize and fit a DeepSurv model
deepsurv_model = DeepSurv()
deepsurv_model.fit(time, event, X)

# Feature importance calculation for neural networks often requires specific techniques
# (e.g., permutation importance, SHAP values) and might not be a direct method.
# Example placeholder:
# feature_importances = deepsurv_model.calculate_feature_importance(X, y)

```

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

### Importing necessary packages

Source: https://github.com/square/pysurvival/blob/master/docs/models/coxph.html

Import all required libraries for data simulation, modeling, and evaluation.

```python
import numpy as np
import pandas as pd
from matplotlib import pyplot as plt
from sklearn.model_selection import train_test_split
from pysurvival.models.simulations import SimulationModel
from pysurvival.models.semi_parametric import CoxPHModel
from pysurvival.utils.metrics import concordance_index
from pysurvival.utils.display import integrated_brier_score
```

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

### Get Survival Function at Specific Times with CoxPH

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Calculates the survival function at specific time points for a given covariate vector using a fitted CoxPH model. This provides detailed survival estimates at discrete times, adjusted for covariates.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import simulate_data
import numpy as np

# Assume coxph_model is already fitted with time, event, X
# time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
# coxph_model.fit(time, event, X)

# Define a specific covariate vector
# X_specific = X[0, :]

# Define specific time points for prediction
# time_points = np.array([10, 50, 100])

# Get the survival function at these time points
# survival_at_times = coxph_model.get_survival_function_at_times(X_specific, time_points)
# print(survival_at_times)

```

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

### Upgrade GCC on MacOS

Source: https://github.com/square/pysurvival/blob/master/docs/installation.html

Use this command to upgrade to the latest version of GCC on MacOS if you have already installed it via Homebrew.

```bash
brew upgrade gcc
```

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

### Get Hazard Function at Specific Times with CoxPH

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Calculates the hazard function at specific time points for a given covariate vector using a fitted CoxPH model. This provides detailed risk progression estimates at discrete times, adjusted for covariates.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import simulate_data
import numpy as np

# Assume coxph_model is already fitted with time, event, X
# time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
# coxph_model.fit(time, event, X)

# Define a specific covariate vector
# X_specific = X[0, :]

# Define specific time points for prediction
# time_points = np.array([10, 50, 100])

# Get the hazard function at these time points
# hazard_at_times = coxph_model.get_hazard_function_at_times(X_specific, time_points)
# print(hazard_at_times)

```

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

### Generating simulation dataset

Source: https://github.com/square/pysurvival/blob/master/docs/models/nonlinear_coxph.html

Initialize a simulation model and generate random samples for training.

```python
# Initializing the simulation model
sim = SimulationModel( survival_distribution = 'weibull',
                       risk_type = 'gaussian',
                       censored_parameter = 2.1,
                       alpha = 0.1, beta=3.2 )

# Generating N random samples 
N = 1000
dataset = sim.generate_data(num_samples = N, num_features=3)

# Showing a few data-points 
dataset.head(2)
```

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

### KernelSVMModel Initialization

Source: https://github.com/square/pysurvival/blob/master/docs/models/kernel_svm.html

Parameters for initializing the KernelSVMModel.

```APIDOC
## KernelSVMModel Parameters

### kernel
- **type**: `str`
- **default**: "gaussian"
- **description**: The type of kernel used to fit the model. Available kernels: Polynomial, Gaussian, Exponential, Tanh, Sigmoid, Rational Quadratic, Inverse Multiquadratic, Multiquadratic.

### scale
- **type**: `float`
- **default**: 1
- **description**: Scale parameter of the kernel function.

### offset
- **type**: `float`
- **default**: 0
- **description**: Offset parameter of the kernel function.

### degree
- **type**: `float`
- **default**: 1
- **description**: Degree parameter of the polynomial/kernel function.
```

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

### Load and Inspect Dataset

Source: https://github.com/square/pysurvival/blob/master/notebooks/Churn Prediction - Predicting when your customers will churn.ipynb

Imports necessary libraries and loads the churn dataset. Prints the shape of the raw dataset and displays the first two rows.

```python
# Importing modules
import pandas as pd
import numpy as np
from matplotlib import pyplot as plt
from pysurvival.datasets import Dataset
%pylab inline

# Reading the dataset
raw_dataset = Dataset('churn').load()
print("The raw_dataset has the following shape: {}.".format(raw_dataset.shape))
raw_dataset.head(2)
```

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

### Preparing training and testing datasets

Source: https://github.com/square/pysurvival/blob/master/docs/models/conditional_survival_forest.html

Splitting the generated data into features and target variables for training.

```python
# Defining the features
features = sim.features

# Building training and testing sets #
index_train, index_test = train_test_split( range(N), test_size = 0.2)
data_train = dataset.loc[index_train].reset_index( drop = True )
data_test  = dataset.loc[index_test].reset_index( drop = True )

# Creating the X, T and E input
X_train, X_test = data_train[features], data_test[features]
T_train, T_test = data_train['time'].values, data_test['time'].values
E_train, E_test = data_train['event'].values, data_test['event'].values
```

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

### Initialize SmoothKaplanMeierModel

Source: https://github.com/square/pysurvival/blob/master/docs/models/smooth_kaplan_meier.html

Initializes the SmoothKaplanMeierModel with specified bandwidth and kernel type. Bandwidth controls smoothing, and kernel defines the smoothing function.

```python
SmoothKaplanMeierModel(bandwidth=0.1, kernel='normal')
```

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

### SimulationModel Instance Creation

Source: https://github.com/square/pysurvival/blob/master/docs/models/simulations.html

How to initialize the SimulationModel class to prepare for generating survival data.

```APIDOC
## SimulationModel Initialization

### Description
Creates an instance of the SimulationModel class to generate random survival times.

### Method
Constructor

### Endpoint
pysurvival.models.simulations.SimulationModel

### Request Example
from pysurvival.models.simulations import SimulationModel
sim = SimulationModel()
```

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

### Initialize and Fit KernelSVMModel

Source: https://github.com/square/pysurvival/blob/master/docs/models/kernel_svm.html

Creates an instance of the KernelSVMModel and fits it to the training data using specified hyperparameters.

```python
svm_model = KernelSVMModel(kernel='Gaussian', scale=0.25)
svm_model.fit(X_train, T_train, E_train, init_method='orthogonal',
    with_bias = True, lr = 0.8,  tol = 1e-3,  l2_reg = 1e-4)
```

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

### Set C++ Compiler Environment Variables on CentOS7

Source: https://github.com/square/pysurvival/blob/master/docs/installation.html

Set the C++ compiler and C compiler paths for CentOS 7. Ensure the version number matches your GCC installation.

```bash
export CXX=/opt/rh/devtoolset-8/root/usr/bin/x86_64-redhat-linux-g++
export CC=/opt/rh/devtoolset-8/root/usr/bin/x86_64-redhat-linux-gcc
```

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

### Import necessary libraries

Source: https://github.com/square/pysurvival/blob/master/docs/models/kernel_svm.html

Import all required libraries for data simulation, model training, and evaluation.

```python
import numpy as np
import pandas as pd
from matplotlib import pyplot as plt
from pysurvival.models.svm import KernelSVMModel
from pysurvival.models.simulations import SimulationModel
from pysurvival.utils.metrics import concordance_index
from sklearn.model_selection import train_test_split
from scipy.stats.stats import pearsonr
```

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

### SmoothKaplanMeierModel Initialization

Source: https://github.com/square/pysurvival/blob/master/docs/models/smooth_kaplan_meier.html

Initializes the SmoothKaplanMeierModel with specified bandwidth and kernel type.

```APIDOC
## SmoothKaplanMeierModel(bandwidth=0.1, kernel='normal')

### Description
Initialize the estimator.

### Parameters
#### Parameters
- **bandwidth** (double) - Optional - controls the degree of the smoothing. The smaller it is the closer to the original KM the function will be, but it will increase the computation time. If it is very large, the resulting model will be smoother than the estimator of KM, but it will stop being as accurate. (default=0.1)
- **kernel** (str) - Optional - defines the type of kernel the model will be using. Here are the possible options:
    - `Uniform`: f(x) = 0 if |x|<1 else f(x) = 0.5
    - `Epanechnikov`: f(x) = 0 if |x| \leq 1 else f(x) = 0.75 \cdot (1 - x^2 )
    - `Normal`: f(x) = \exp( -x^2/2) / \sqrt{2 \cdot \pi}
    - `Biweight`: f(x) = 0 if |x| \leq 1 else f(x)=(15/16) \cdot (1-x^2)^2
    - `Triweight`: f(x) = 0 if |x| \leq 1 else f(x)=(35/32) \cdot (1-x^2)^3
    - `Cosine`: f(x) = 0 if |x| \leq 1 else f(x)=(\pi/4) \cdot \cos( \pi \cdot x/2. ) (default='normal')
```

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

### Load Built-in Datasets

Source: https://context7.com/square/pysurvival/llms.txt

Use the Dataset class to load tutorial datasets with automatic train/test splitting and feature extraction. Available datasets include 'churn', 'maintenance', 'credit_risk', 'employee_attrition', and 'simple_example'.

```python
from pysurvival.datasets import Dataset

# Load a dataset - available options: 'simple_example', 'maintenance',
# 'credit_risk', 'employee_attrition', 'churn'
dataset = Dataset('churn')

# Load and automatically split into train/test sets (30% test by default)
X_train, T_train, E_train, X_test, T_test, E_test = dataset.load_train_test(test_size=0.3)

# Or load the full dataset as a DataFrame
data = dataset.load()
print(f"Dataset: {dataset.name}")
print(f"Time column: {dataset.time_column}")
print(f"Event column: {dataset.event_column}")
print(f"Features: {dataset.features}")
```

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

### Get Feature Importance for Survival Forest

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Retrieves the feature importance scores from a fitted Survival Forest model. This helps identify which features are most influential in predicting survival outcomes.

```python
from pysurvival.models.non_parametric import SurvivalForest
from pysurvival.utils.data import simulate_data

# Simulate survival data
time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)

# Initialize and fit a Survival Forest model
survival_forest = SurvivalForest()
survival_forest.fit(time, event, X)

# Get feature importances
# feature_importances = survival_forest.feature_importance
# print(feature_importances)

```

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

### Split dataset into training and testing sets

Source: https://github.com/square/pysurvival/blob/master/docs/models/kernel_svm.html

Prepare the dataset for model training by defining features and splitting the data into training and testing subsets. This is a standard practice for model evaluation.

```python
# Defining the features
features = sim.features
```

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

### NeuralMultiTaskModel Constructor

Source: https://github.com/square/pysurvival/blob/master/docs/models/neural_mtlr.html

Initializes an instance of the NeuralMultiTaskModel.

```APIDOC
## NeuralMultiTaskModel Constructor

### Description
Initializes an instance of the NeuralMultiTaskModel.

### Method
`__init__`

### Endpoint
`pysurvival.models.multi_task.NeuralMultiTaskModel`

### Parameters
#### Request Body
- **structure** (list of dictionaries) - Required - Provides the structure of the MLP built within the N-MTLR. Example: `[{'activation': 'ReLU', 'num_units': 128}]`.
- **bins** (int) - Optional - Number of subdivisions of the time axis (default=100).
- **auto_scaler** (boolean) - Optional - Determines whether a sklearn scaler should be automatically applied (default=True).
```

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

### Load and inspect dataset

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.ipynb

Imports necessary libraries and loads the maintenance dataset for initial inspection.

```python
# Importing modules
import pandas as pd
import numpy as np
from matplotlib import pyplot as plt
%pylab inline

# Reading the dataset
raw_dataset = Dataset('maintenance').load()
print("The raw_dataset has the following shape: {}.".format(raw_dataset.shape))
raw_dataset.head(3)
```

```python
raw_dataset.info()
```

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

### Importing necessary libraries

Source: https://github.com/square/pysurvival/blob/master/docs/models/linear_svm.html

Import all required libraries for data manipulation, modeling, and visualization.

```python
import numpy as np
import pandas as pd
from matplotlib import pyplot as plt
from pysurvival.models.svm import LinearSVMModel
from pysurvival.models.simulations import SimulationModel
from pysurvival.utils.metrics import concordance_index
from sklearn.model_selection import train_test_split
from scipy.stats.stats import pearsonr
# %pylab inline # to use in jupyter notebooks
```

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

### Set C++ Compiler Environment Variables on MacOS

Source: https://github.com/square/pysurvival/blob/master/docs/installation.html

Assign the C++ compiler and C compiler paths to the CC and CXX environment variables. Adjust the version number to match your GCC installation.

```bash
export CXX=/usr/local/Cellar/gcc/8.2.0/bin/g++-8
export CC=/usr/local/Cellar/gcc/8.2.0/bin/gcc-8
```

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

### Initializing and fitting Linear SVM model

Source: https://github.com/square/pysurvival/blob/master/docs/models/linear_svm.html

Create an instance of LinearSVMModel and fit it to the training data using specified initialization method.

```python
svm_model = LinearSVMModel()
svm_model.fit(X_train, T_train, E_train, init_method='he_uniform',
```

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

### Initialize NeuralMultiTaskModel

Source: https://github.com/square/pysurvival/blob/master/docs/models/neural_mtlr.html

Instantiate the NeuralMultiTaskModel with a specified structure for the MLP. The structure defines the layers and activation functions. Optional parameters include the number of time bins and whether to use an auto-scaler.

```python
structure = [ {'activation': 'ReLU', 'num_units': 128}, ]
model = NeuralMultiTaskModel(structure=structure)
```

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

### Build and Train CoxPH and MTLR Models

Source: https://github.com/square/pysurvival/blob/master/docs/index.html

This snippet demonstrates loading data, building a CoxPH model and a LinearMultiTaskModel, training them, and evaluating their performance using concordance index. Ensure necessary modules are imported.

```python
# Loading the modules
from pysurvival.models.semi_parametric import CoxPHModel
from pysurvival.models.multi_task import LinearMultiTaskModel
from pysurvival.datasets import Dataset
from pysurvival.utils.metrics import concordance_index

# Loading and splitting a simple example into train/test sets
X_train, T_train, E_train, X_test, T_test, E_test \
    = Dataset('simple_example').load_train_test()

# Building a CoxPH model
coxph_model = CoxPHModel()
coxph_model.fit(X=X_train, T=T_train, E=E_train, init_method='he_uniform',
                l2_reg = 1e-4, lr = .4, tol = 1e-4)

# Building a MTLR model
mtlr = LinearMultiTaskModel()
mtlr.fit(X=X_train, T=T_train, E=E_train, init_method = 'glorot_uniform',
         optimizer ='adam', lr = 8e-4)

# Checking the model performance
c_index1 = concordance_index(model=coxph_model, X=X_test, T=T_test, E=E_test )
print("CoxPH model c-index = {:.2f}".format(c_index1))

c_index2 = concordance_index(model=mtlr, X=X_test, T=T_test, E=E_test )
print("MTLR model c-index = {:.2f}".format(c_index2))
```

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

### Initialize Survival Model with Custom Parameters

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Initializes a survival model with specific parameters, allowing for fine-tuning or setting initial values.

```python
from pysurvival.models.parametric import Weibull

# Initialize a Weibull model with custom parameters (e.g., shape=1.5, scale=100)
# weibull_model = Weibull(alpha=1.5, beta=100.0)

```

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

### Get Feature Importance for CoxPH

Source: https://github.com/square/pysurvival/blob/master/notebooks/Predictive Maintenance - Predicting when a machine will break.html

Feature importance for CoxPH models is typically derived from the magnitude and significance of the estimated coefficients. Larger absolute coefficient values suggest a stronger impact on the hazard rate.

```python
from pysurvival.models.coxph import CoxPH
from pysurvival.utils.data import simulate_data

# Assume coxph_model is already fitted with time, event, X
# time, event, X = simulate_data(n_samples=100, n_features=5, n_informative=3)
# coxph_model.fit(time, event, X)

# Feature importance can be inferred from coxph_model.hazard_ratios
# and their associated p-values from coxph_model.summary()
# Example:
# hazard_ratios = coxph_model.hazard_ratios
# print("Hazard Ratios (Feature Importance proxy):")
# print(hazard_ratios)

```

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

### Generating simulation data

Source: https://github.com/square/pysurvival/blob/master/docs/models/neural_mtlr.html

Initializes a simulation model and generates a dataset for training.

```python
# Initializing the simulation model
sim = SimulationModel( survival_distribution = 'Log-Logistic',
                       risk_type = 'gaussian',
                       censored_parameter = 10.1,
                       alpha = 0.1, beta=3.2 )

# Generating N random samples 
N = 1000
dataset = sim.generate_data(num_samples = N, num_features = 3)

# Showing a few data-points 
dataset.head(2)
```

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

### Initialize RandomSurvivalForestModel

Source: https://github.com/square/pysurvival/blob/master/docs/models/random_survival_forest.html

Instantiate the Random Survival Forest model. The `num_trees` parameter controls the number of trees in the forest.

```python
RandomSurvivalForestModel(num_trees = 10)
```
Tokens:
Raw