Survival regression

Often we have additional data aside from the duration that we want to use. The technique is called survival regression – the name implies we regress covariates (e.g., age, country, etc.) against another variable – in this case durations. Similar to the logic in the first part of this tutorial, we cannot use traditional methods like linear regression because of censoring.

There are a few popular models in survival regression: Cox’s model, accelerated failure models, and Aalen’s additive model. All models attempt to represent the hazard rate \(h(t | x)\) as a function of \(t\) and some covariates \(x\). We explore these models next.

The dataset for regression

The dataset required for survival regression must be in the format of a Pandas DataFrame. Each row of the DataFrame should be an observation. There should be a column denoting the durations of the observations. There may be a column denoting the event status of each observation (1 if event occurred, 0 if censored). There are also the additional covariates you wish to regress against. Optionally, there could be columns in the DataFrame that are used for stratification, weights, and clusters which will be discussed later in this tutorial.

An example dataset we will use is the Rossi recidivism dataset, available in lifelines as load_rossi().

from lifelines.datasets import load_rossi

rossi = load_rossi()

     week  arrest  fin  age  race  wexp  mar  paro  prio
0      20       1    0   27     1     0    0     1     3
1      17       1    0   18     1     0    0     1     8
2      25       1    0   19     0     1    0     1    13
3      52       0    1   23     1     1    1     1     1

The dataframe rossi contains 432 observations. The week column is the duration, the arrest column is the event occurred, and the other columns represent variables we wish to regress against.

If you need to first clean or transform your dataset (encode categorical variables, add interaction terms, etc.), that should happen before using lifelines. Libraries like Pandas and Patsy help with that.

Cox’s proportional hazard model

The idea behind Cox’s proportional hazard model model is that the log-hazard of an individual is a linear function of their static covariates and a population-level baseline hazard that changes over time. Mathematically:

\[\underbrace{h(t | x)}_{\text{hazard}} = \overbrace{b_0(t)}^{\text{baseline hazard}} \underbrace{\exp \overbrace{\left(\sum_{i=1}^n b_i (x_i - \overline{x_i})\right)}^{\text{log-partial hazard}}}_ {\text{partial hazard}}\]

Note a few facts about this model: the only time component is in the baseline hazard, \(b_0(t)\). In the above product, the partial hazard is a time-invariant scalar factor that only increases or decreases the baseline hazard. Thus a changes in covariates will only increase or decrease the baseline hazard.


In other regression models, a column of 1s might be added that represents that intercept or baseline. This is not necessary in the Cox model. In fact, there is no intercept in the additive Cox model - the baseline hazard represents this. lifelines will will throw warnings and may experience convergence errors if a column of 1s is present in your dataset.

Running the regression

The implementation of the Cox model in lifelines is under CoxPHFitter. Like R, it has a print_summary() function that prints a tabular view of coefficients and related stats.

from lifelines import CoxPHFitter
from lifelines.datasets import load_rossi

rossi_dataset = load_rossi()

cph = CoxPHFitter(), duration_col='week', event_col='arrest', show_progress=True)

cph.print_summary()  # access the results using cph.summary

<lifelines.CoxPHFitter: fitted with 432 observations, 318 censored>
      duration col = 'week'
         event col = 'arrest'
number of subjects = 432
  number of events = 114
    log-likelihood = -658.75
  time fit was run = 2019-01-27 23:10:15 UTC

      coef  exp(coef)  se(coef)     z      p  -log2(p)  lower 0.95  upper 0.95
fin  -0.38       0.68      0.19 -1.98   0.05      4.40       -0.75       -0.00
age  -0.06       0.94      0.02 -2.61   0.01      6.79       -0.10       -0.01
race  0.31       1.37      0.31  1.02   0.31      1.70       -0.29        0.92
wexp -0.15       0.86      0.21 -0.71   0.48      1.06       -0.57        0.27
mar  -0.43       0.65      0.38 -1.14   0.26      1.97       -1.18        0.31
paro -0.08       0.92      0.20 -0.43   0.66      0.59       -0.47        0.30
prio  0.09       1.10      0.03  3.19 <0.005      9.48        0.04        0.15
Concordance = 0.64
Likelihood ratio test = 33.27 on 7 df, -log2(p)=15.37

To access the coefficients and the baseline hazard directly, you can use params_ and baseline_hazard_ respectively. Taking a look at these coefficients for a moment, prio (the number of prior arrests) has a coefficient of about 0.09. Thus, a one unit increase in prio means the the baseline hazard will increase by a factor of \(\exp{(0.09)} = 1.10\) - about a 10% increase. Recall, in the Cox proportional hazard model, a higher hazard means more at risk of the event occurring. The value \(\exp{(0.09)}\) is called the hazard ratio, a name that will be clear with another example.

Consider the coefficient of mar (whether the subject is married or not). The values in the column are binary: 0 or 1, representing either not married or married. The value of the coefficient associated with mar, \(\exp{(-.43)}\), is the value of ratio of hazards associated with being married, that is:

\[\exp(-0.43) = \frac{\text{hazard of married subjects at time $t$}}{\text{hazard of unmarried subjects at time $t$}}\]

Note that left-hand side is a constant (specifically, it’s independent of time, \(t\)), but the right-hand side has two factors that can vary with time. The proportional assumption is that this is true in reality. That is, hazards can change over time, but their ratio between levels remains a constant. Later we will deal with checking this assumption.


Fitting the Cox model to the data involves using iterative methods. lifelines takes extra effort to help with convergence, so please be attentive to any warnings that appear. Fixing any warnings will generally help convergence and decrease the number of iterative steps required. If you wish to see the fitting, there is a show_progress parameter in fit() function. For further help, see Problems with convergence in the Cox proportional hazard model.

After fitting, the value of the maximum log-likelihood this available using log_likelihood. The variance matrix of the coefficients is available under variance_matrix_.

Goodness of fit

After fitting, you may want to know how “good” of a fit your model was to the data. A few methods the author has found useful is to


After fitting, you can use use the suite of prediction methods: predict_partial_hazard(), predict_survival_function(), etc.

X = rossi_dataset


cph.predict_survival_function(X, times=[5., 25., 50.])


A common use case is to predict the event time of censored subjects. This is easy to do, but we first have to calculate an important conditional probability. Let \(T\) be the (random) event time for some subject, and \(S(t)≔P(T > t)\) be their survival function. We are interested to know What is the new survival function, given I know the subject has lived past time s, where s < t? Mathematically:

\[\begin{split}\begin{align*} P(T > t \;|\; T > s) &= \frac{P(T > t \;\text{and}\; T > s)}{P(T > s)} \\ &= \frac{P(T > t)}{P(T > s)} \\ &= \frac{S(t)}{S(s)} \end{align*}\end{split}\]

Thus we scale the original survival function by the survival function at time \(s\) (everything prior to \(s\) should be mapped to 1.0 as well, since we are working with probabilities and we know that the subject was alive before \(s\)).

Back to our original problem of predicting the event time of censored individuals, lifelines has all this math and logic built in when using the conditional_after kwarg.

censored_subjects = rossi.loc[~rossi['arrest'].astype(bool)]
censored_subjects_last_obs = censored_subjects['week']

cph.predict_partial_hazard(censored_subjects, conditional_after=censored_subjects_last_obs)

cph.predict_survival_function(censored_subjects, times=[5., 25., 50.], conditional_after=censored_subjects_last_obs)

cph.predict_median(censored_subjects, conditional_after=censored_subjects_last_obs)

Plotting the coefficients

With a fitted model, an alternative way to view the coefficients and their ranges is to use the plot method.

from lifelines.datasets import load_rossi
from lifelines import CoxPHFitter

rossi_dataset = load_rossi()
cph = CoxPHFitter(), duration_col='week', event_col='arrest', show_progress=True)


Plotting the effect of varying a covariate

After fitting, we can plot what the survival curves look like as we vary a single covariate while holding everything else equal. This is useful to understand the impact of a covariate, given the model. To do this, we use the plot_covariate_groups() method and give it the covariate of interest, and the values to display.

from lifelines.datasets import load_rossi
from lifelines import CoxPHFitter

rossi_dataset = load_rossi()
cph = CoxPHFitter(), duration_col='week', event_col='arrest', show_progress=True)

cph.plot_covariate_groups('prio', [0, 2, 4, 6, 8, 10], cmap='coolwarm')

The plot_covariate_groups() method can accept multiple covariates as well. This is useful for two purposes:

  1. There are derivative features in your dataset. For example, suppose you have included year and year**2 in your dataset. It doesn’t make sense to just vary year and leave year**2 fixed. You’ll need to specify manually the values the covariates take on in a N-d array or list (where N is the number of covariates being varied.)
    ['year', 'year**2'],
        [0, 0],
        [1, 1],
        [2, 4],
        [3, 9],
        [8, 64],
  1. This feature is also useful for analyzing categorical variables. In your regression, you may have dummy variables (also called one-hot-encoded variables) in your DataFrame that represent some categorical variable. To simultaneously plot the survival curves of each category, all else being equal, we can use:
    ['d1', 'd2' 'd3', 'd4', 'd5'],

The reason why we use np.eye is because we want each row of the matrix to “turn on” one category and “turn off” the others.

Checking the proportional hazards assumption

CoxPHFitter has a check_assumptions() method that will output violations of the proportional hazard assumption. For a tutorial on how to fix violations, see Testing the Proportional Hazard Assumptions.

Non-proportional hazards is a case of model misspecification. Suggestions are to look for ways to stratify a column (see docs below), or use a time varying model.


Sometimes one or more covariates may not obey the proportional hazard assumption. In this case, we can allow the covariate(s) to still be including in the model without estimating its effect. This is called stratification. At a high level, think of it as splitting the dataset into N smaller datasets, defined by the unique values of the stratifying covariate(s). Each dataset has its own baseline hazard (the non-parametric part of the model), but they all share the regression parameters (the parametric part of the model). Since covariates are the same within each dataset, there is no regression parameter for the covariates stratified on, hence they will not show up in the output. However there will be N baseline hazards under baseline_cumulative_hazard_.

To specify variables to be used in stratification, we define them in the call to fit():

from lifelines.datasets import load_rossi
from lifelines import CoxPHFitter

rossi_dataset = load_rossi()
cph = CoxPHFitter(), 'week', event_col='arrest', strata=['race'], show_progress=True)

cph.print_summary()  # access the results using cph.summary

<lifelines.CoxPHFitter: fitted with 432 observations, 318 censored>
      duration col = 'week'
         event col = 'arrest'
            strata = ['race']
number of subjects = 432
  number of events = 114
    log-likelihood = -620.56
  time fit was run = 2019-01-27 23:08:35 UTC

      coef  exp(coef)  se(coef)     z      p  -log2(p)  lower 0.95  upper 0.95
fin  -0.38       0.68      0.19 -1.98   0.05      4.39       -0.75       -0.00
age  -0.06       0.94      0.02 -2.62   0.01      6.83       -0.10       -0.01
wexp -0.14       0.87      0.21 -0.67   0.50      0.99       -0.56        0.27
mar  -0.44       0.64      0.38 -1.15   0.25      2.00       -1.19        0.31
paro -0.09       0.92      0.20 -0.44   0.66      0.60       -0.47        0.30
prio  0.09       1.10      0.03  3.21 <0.005      9.56        0.04        0.15
Concordance = 0.64
Likelihood ratio test = 109.63 on 6 df, -log2(p)=68.48

# (49, 2)

Weights & robust errors

Observations can come with weights, as well. These weights may be integer values representing some commonly occurring observation, or they may be float values representing some sampling weights (ex: inverse probability weights). In the fit() method, an kwarg is present for specifying which column in the DataFrame should be used as weights, ex: CoxPHFitter(df, 'T', 'E', weights_col='weights').

When using sampling weights, it’s correct to also change the standard error calculations. That is done by turning on the robust flag in fit(). Internally, CoxPHFitter will use the sandwich estimator to compute the errors.

from lifelines import CoxPHFitter

df = pd.DataFrame({
    'T': [5, 3, 9, 8, 7, 4, 4, 3, 2, 5, 6, 7],
    'E': [1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 0],
    'weights': [1.1, 0.5, 2.0, 1.6, 1.2, 4.3, 1.4, 4.5, 3.0, 3.2, 0.4, 6.2],
    'month': [10, 3, 9, 8, 7, 4, 4, 3, 2, 5, 6, 7],
    'age': [4, 3, 9, 8, 7, 4, 4, 3, 2, 5, 6, 7],

cph = CoxPHFitter(), 'T', 'E', weights_col='weights', robust=True)

See more examples in Adding weights to observations in a Cox model.

Clusters & correlations

Another property your dataset may have is groups of related subjects. This could be caused by:

  • a single individual having multiple occurrences, and hence showing up in the dataset more than once.
  • subjects that share some common property, like members of the same family or being matched on propensity scores.

We call these grouped subjects “clusters”, and assume they are designated by some column in the DataFrame (example below). When using cluster, the point estimates of the model don’t change, but the standard errors will increase. An intuitive argument for this is that 100 observations on 100 individuals provide more information than 100 observations on 10 individuals (or clusters).

from lifelines import CoxPHFitter

df = pd.DataFrame({
    'T': [5, 3, 9, 8, 7, 4, 4, 3, 2, 5, 6, 7],
    'E': [1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 0],
    'month': [10, 3, 9, 8, 7, 4, 4, 3, 2, 5, 6, 7],
    'age': [4, 3, 9, 8, 7, 4, 4, 3, 2, 5, 6, 7],
    'id': [1, 1, 1, 1, 2, 3, 3, 4, 4, 5, 6, 7]

cph = CoxPHFitter(), 'T', 'E', cluster_col='id')

For more examples, see Correlations between subjects in a Cox model.


After fitting a Cox model, we can look back and compute important model residuals. These residuals can tell us about non-linearities not captured, violations of proportional hazards, and help us answer other useful modeling questions. See Assessing Cox model fit using residuals.

Accelerated failure time models

Suppose we have two populations, A and B, with different survival functions, \(S_A(t)\) and \(S_B(t)\), and they are related by some accelerated failure rate, \(\lambda\):

\[S_A(t) = S_B\left(\frac{t}{\lambda}\right)\]

This can be interpreted as slowing down or speeding up moving along the survival function. A classic example of this is that dogs age at 7 times the rate of humans, i.e. \(\lambda = \frac{1}{7}\). This model has some other nice properties: the average survival time of population B is \({\lambda}\) times the average survival time of population A. Likewise with the median survival time.

More generally, we can model the \(\lambda\) as a function of covariates available, that is:

\[\begin{split}S_A(t) = S_B\left(\frac{t}{\lambda(x)}\right)\\ \lambda(x) = \exp\left(b_0 + \sum_{i=1}^n b_i x_i \right)\end{split}\]

This model can accelerate or decelerate failure times depending on subjects’ covariates. Another nice feature of this is the ease of interpretation of the coefficients: a unit increase in \(x_i\) means the average/median survival time changes by a factor of \(\exp(b_i)\).


An important note on interpretation: Suppose \(b_i\) was positive, then the factor \(\exp(b_i)\) is greater than 1, which will decelerate the event time since we divide time by the factor ⇿ increase mean/median survival. Hence, it will be a protective effect. Likewise, a negative \(b_i\) will hasten the event time ⇿ reduce the mean/median survival time. This interpretation is opposite of how the sign influences event times in the Cox model! This is standard survival analysis convention.

Next, we pick a parametric form for the survival function, \(S(t)\). The most common is the Weibull form. So if we assume the relationship above and a Weibull form, our hazard function is quite easy to write down:

\[H(t; x) = \left( \frac{t}{\lambda(x)} \right)^\rho\]

We call these accelerated failure time models, shortened often to just AFT models. Using lifelines, we can fit this model (and the unknown \(\rho\) parameter too).

The Weibull AFT model

The Weibull AFT model is implemented under WeibullAFTFitter. The API for the class is similar to the other regression models in lifelines. After fitting, the coefficients can be accessed using params_ or summary, or alternatively printed using print_summary().

from lifelines import WeibullAFTFitter
from lifelines.datasets import load_rossi

rossi_dataset = load_rossi()

aft = WeibullAFTFitter(), duration_col='week', event_col='arrest')

aft.print_summary(3)  # access the results using aft.summary

<lifelines.WeibullAFTFitter: fitted with 432 observations, 318 censored>
      duration col = 'week'
         event col = 'arrest'
number of subjects = 432
  number of events = 114
    log-likelihood = -679.917
  time fit was run = 2019-02-20 17:47:19 UTC

                     coef  exp(coef)  se(coef)      z       p  -log2(p)  lower 0.95  upper 0.95
lambda_ fin         0.272      1.313     0.138  1.973   0.049     4.365       0.002       0.543
        age         0.041      1.042     0.016  2.544   0.011     6.512       0.009       0.072
        race       -0.225      0.799     0.220 -1.021   0.307     1.703      -0.656       0.207
        wexp        0.107      1.112     0.152  0.703   0.482     1.053      -0.190       0.404
        mar         0.311      1.365     0.273  1.139   0.255     1.973      -0.224       0.847
        paro        0.059      1.061     0.140  0.421   0.674     0.570      -0.215       0.333
        prio       -0.066      0.936     0.021 -3.143   0.002     9.224      -0.107      -0.025
        _intercept  3.990     54.062     0.419  9.521 <0.0005    68.979       3.169       4.812
rho_    _intercept  0.339      1.404     0.089  3.809 <0.0005    12.808       0.165       0.514
Concordance = 0.640
Log-likelihood ratio test = 33.416 on 7 df, -log2(p)=15.462

From above, we can see that prio, which is the number of previous incarcerations, has a large negative coefficient. This means that each addition incarcerations changes a subject’s mean/median survival time by \(\exp(-0.066) = 0.936\), approximately a 7% decrease in mean/median survival time. What is the mean/median survival time?


# 100.325
# 118.67

What does the rho_    _intercept row mean in the above table? Internally, we model the log of the rho_ parameter, so the value of \(\rho\) is the exponential of the value, so in case above it’s \(\hat{\rho} = \exp0.339 = 1.404\). This brings us to the next point - modelling \(\rho\) with covariates as well:

Modeling ancillary parameters

In the above model, we left the parameter \(\rho\) as a single unknown. We can also choose to model this parameter as well. Why might we want to do this? It can help in survival prediction to allow heterogeneity in the \(\rho\) parameter. The model is no longer an AFT model, but we can still recover and understand the influence of changing a covariate by looking at its outcome plot (see section below). To model \(\rho\), we use the ancillary_df keyword argument in the call to fit(). There are four valid options:

  1. False or None: explicitly do not model the rho_ parameter (except for its intercept).
  2. a Pandas DataFrame. This option will use the columns in the Pandas DataFrame as the covariates in the regression for rho_. This DataFrame could be a equal to, or a subset of, the original dataset using for modeling lambda_, or it could be a totally different dataset.
  3. True. Passing in True will internally reuse the dataset that is being used to model lambda_.
aft = WeibullAFTFitter(), duration_col='week', event_col='arrest', ancillary_df=False)
# identical to, duration_col='week', event_col='arrest', ancillary_df=None), duration_col='week', event_col='arrest', ancillary_df=some_df), duration_col='week', event_col='arrest', ancillary_df=True)
# identical to, duration_col='week', event_col='arrest', ancillary_df=rossi)


<lifelines.WeibullAFTFitter: fitted with 432 observations, 318 censored>
      duration col = 'week'
         event col = 'arrest'
number of subjects = 432
  number of events = 114
    log-likelihood = -669.40
  time fit was run = 2019-02-20 17:42:55 UTC

                    coef  exp(coef)  se(coef)     z      p  -log2(p)  lower 0.95  upper 0.95
lambda_ fin         0.24       1.28      0.15  1.60   0.11      3.18       -0.06        0.55
        age         0.10       1.10      0.03  3.43 <0.005     10.69        0.04        0.16
        race        0.07       1.07      0.19  0.36   0.72      0.48       -0.30        0.44
        wexp       -0.34       0.71      0.15 -2.22   0.03      5.26       -0.64       -0.04
        mar         0.26       1.30      0.30  0.86   0.39      1.35       -0.33        0.85
        paro        0.09       1.10      0.15  0.61   0.54      0.88       -0.21        0.39
        prio       -0.08       0.92      0.02 -4.24 <0.005     15.46       -0.12       -0.04
        _intercept  2.68      14.65      0.60  4.50 <0.005     17.14        1.51        3.85
rho_    fin        -0.01       0.99      0.15 -0.09   0.92      0.11       -0.31        0.29
        age        -0.05       0.95      0.02 -3.10 <0.005      9.01       -0.08       -0.02
        race       -0.46       0.63      0.25 -1.79   0.07      3.77       -0.95        0.04
        wexp        0.56       1.74      0.17  3.32 <0.005     10.13        0.23        0.88
        mar         0.10       1.10      0.27  0.36   0.72      0.47       -0.44        0.63
        paro        0.02       1.02      0.16  0.12   0.90      0.15       -0.29        0.33
        prio        0.03       1.03      0.02  1.44   0.15      2.73       -0.01        0.08
        _intercept  1.48       4.41      0.41  3.60 <0.005     11.62        0.68        2.29
Concordance = 0.63
Log-likelihood ratio test = 54.45 on 14 df, -log2(p)=19.83


The plotting API is the same as in CoxPHFitter. We can view all covariates in a forest plot:

wft = WeibullAFTFitter().fit(rossi, 'week', 'arrest', ancillary_df=True)

We can observe the influence a variable in the model by plotting the outcome (i.e. survival) of changing the variable. This is done using plot_covariate_groups(), and this is also a nice time to observe the effects of modeling rho_ vs keeping it fixed. Below we fit the Weibull model to the same dataset twice, but in the first model we model rho_ and in the second model we don’t. We when vary the prio (which is the number of prior arrests) and observe how the survival changes.

fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(10, 4))

times = np.arange(0, 100)
wft_model_rho = WeibullAFTFitter().fit(rossi, 'week', 'arrest', ancillary_df=True, timeline=times)
wft_model_rho.plot_covariate_groups('prio', range(0, 16, 3), cmap='coolwarm', ax=ax[0])
ax[0].set_title("Modelling rho_")

wft_not_model_rho = WeibullAFTFitter().fit(rossi, 'week', 'arrest', ancillary_df=False, timeline=times)
wft_not_model_rho.plot_covariate_groups('prio', range(0, 16, 3), cmap='coolwarm', ax=ax[1])
ax[1].set_title("Not modelling rho_");

Comparing a few of these survival functions side by side:

fig, ax = plt.subplots(nrows=1, ncols=1, figsize=(7, 4))

wft_model_rho.plot_covariate_groups('prio', range(0, 16, 5), cmap='coolwarm', ax=ax, lw=2, plot_baseline=False)
wft_not_model_rho.plot_covariate_groups('prio', range(0, 16, 5), cmap='coolwarm', ax=ax, ls='--', lw=2, plot_baseline=False)

You read more about and see other examples of the extensions to plot_covariate_groups()


Given a new subject, we ask questions about their future survival? When are they likely to experience the event? What does their survival function look like? The WeibullAFTFitter is able to answer these. If we have modeled the ancillary covariates, we are required to include those as well:

X = rossi.loc[:10]

aft.predict_cumulative_hazard(X, ancillary_df=X)
aft.predict_survival_function(X, ancillary_df=X)
aft.predict_median(X, ancillary_df=X)
aft.predict_percentile(X, p=0.9, ancillary_df=X)
aft.predict_expectation(X, ancillary_df=X)

When predicting time remaining for uncensored individuals, you can use the conditional_after kwarg:

censored_X = rossi.loc[~rossi['arrest'].astype(bool)]
censored_subjects_last_obs = censored_X['week']

aft.predict_cumulative_hazard(censored_X, ancillary_df=censored_X, conditional_after=censored_subjects_last_obs)
aft.predict_survival_function(censored_X, ancillary_df=censored_X, conditional_after=censored_subjects_last_obs)
aft.predict_median(censored_X, ancillary_df=censored_X, conditional_after=censored_subjects_last_obs)
aft.predict_percentile(X, p=0.9, ancillary_df=censored_X, conditional_after=censored_subjects_last_obs)

There are two hyper-parameters that can be used to to achieve a better test score. These are penalizer and l1_ratio in the call to WeibullAFTFitter. The penalizer is similar to scikit-learn’s ElasticNet model, see their docs.

aft_with_elastic_penalty = WeibullAFTFitter(penalizer=4.0, l1_ratio=1.0), 'week', 'arrest')


<lifelines.WeibullAFTFitter: fitted with 432 observations, 318 censored>
      duration col = 'week'
         event col = 'arrest'
         penalizer = 4.0
          l1_ratio = 1.0
number of subjects = 432
  number of events = 114
    log-likelihood = -2710.95
  time fit was run = 2019-02-20 19:53:29 UTC

                    coef  exp(coef)  se(coef)     z      p  -log2(p)  lower 0.95  upper 0.95
lambda_ fin         0.00       1.00      0.08  0.00   1.00      0.00       -0.15        0.15
        age         0.13       1.14      0.01 12.27 <0.005    112.47        0.11        0.15
        race        0.55       1.73      0.09  5.80 <0.005     27.16        0.36        0.73
        wexp        0.00       1.00      0.09  0.00   1.00      0.00       -0.17        0.17
        mar         0.00       1.00      0.14  0.01   0.99      0.01       -0.27        0.28
        paro        0.00       1.00      0.08  0.01   0.99      0.01       -0.16        0.16
        prio        0.00       1.00      0.01  0.00   1.00      0.00       -0.03        0.03
        _intercept  0.00       1.00      0.19  0.00   1.00      0.00       -0.38        0.38
rho_    _intercept -0.00       1.00       nan   nan    nan       nan         nan         nan
Concordance = 0.60
Log-likelihood ratio test = -4028.65 on 7 df, -log2(p)=-0.00

The Log-Normal and Log-Logistic AFT model

There are also the LogNormalAFTFitter and LogLogisticAFTFitter models, which instead of assuming that the survival time distribution is Weibull, we assume it is Log-Normal or Log-Logistic, respectively. They have identical APIs to the WeibullAFTFitter, but the parameter names are different.

from lifelines import LogLogisticAFTFitter
from lifelines import LogNormalAFTFitter

llf = LogLogisticAFTFitter().fit(rossi, 'week', 'arrest')
lnf = LogNormalAFTFitter().fit(rossi, 'week', 'arrest')

Model selection for AFT models

Often, you don’t know a priori which AFT model to use. Each model has some assumptions built-in (not implemented yet in lifelines), but a quick and effective method is to compare the log-likelihoods for each fitted model. (Technically, we are comparing the AIC, but the number of parameters for each model is the same, so we can simply and just look at the log-likelihood). Generally, given the same dataset and number of parameters, a better fitting model has a larger log-likelihood. We can look at the log-likelihood for each fitted model and select the largest one.

from lifelines import LogLogisticAFTFitter, WeibullAFTFitter, LogNormalAFTFitter
from lifelines.datasets import load_rossi

rossi = load_rossi()

llf = LogLogisticAFTFitter().fit(rossi, 'week', 'arrest')
lnf = LogNormalAFTFitter().fit(rossi, 'week', 'arrest')
wf = WeibullAFTFitter().fit(rossi, 'week', 'arrest')

print(llf.log_likelihood_)  # -679.938
print(lnf.log_likelihood_)  # -683.234
print(wf.log_likelihood_)   # -679.916, slightly the best model.

# with some heterogeneity in the ancillary parameters
ancillary_df = rossi[['prio']]
llf = LogLogisticAFTFitter().fit(rossi, 'week', 'arrest', ancillary_df=ancillary_df)
lnf = LogNormalAFTFitter().fit(rossi, 'week', 'arrest', ancillary_df=ancillary_df)
wf = WeibullAFTFitter().fit(rossi, 'week', 'arrest', ancillary_df=ancillary_df)

print(llf.log_likelihood_) # -678.94, slightly the best model.
print(lnf.log_likelihood_) # -680.39
print(wf.log_likelihood_)  # -679.60

Left, right and interval censored data

The AFT models have APIs that handle left and interval censored data, too. The API for them is different than the API for fitting to right censored data. Here’s an example with interval censored data.

from lifelines.datasets import load_diabetes

df = load_diabetes()
df['gender'] = df['gender'] == 'male'

   left  right  gender
1    24     27    True
2    22     22   False
3    37     39    True
4    20     20    True
5     1     16    True

wf = WeibullAFTFitter().fit_interval_censoring(df, lower_bound_col='left', upper_bound_col='right')

<lifelines.WeibullAFTFitter: fitted with 731 observations, 136 censored>
         event col = 'E'
number of subjects = 731
  number of events = 595
    log-likelihood = -2027.20
  time fit was run = 2019-04-11 19:39:42 UTC

                    coef exp(coef)  se(coef)      z      p  -log2(p)  lower 0.95  upper 0.95
lambda_ gender      0.05      1.05      0.03   1.66   0.10      3.38       -0.01        0.10
        _intercept  2.91     18.32      0.02 130.15 <0.005       inf        2.86        2.95
rho_    _intercept  1.04      2.83      0.03  36.91 <0.005    988.46        0.98        1.09
Log-likelihood ratio test = 2.74 on 1 df, -log2(p)=3.35

Another example of using lifelines for interval censored data is located here.

Aalen’s additive model


This implementation is still experimental.

Aalen’s Additive model is another regression model we can use. Like the Cox model, it defines the hazard rate, but instead of the linear model being multiplicative like the Cox model, the Aalen model is additive. Specifically:

\[h(t|x) = b_0(t) + b_1(t) x_1 + ... + b_N(t) x_N\]

Inference typically does not estimate the individual \(b_i(t)\) but instead estimates \(\int_0^t b_i(s) \; ds\) (similar to the estimate of the hazard rate using NelsonAalenFitter). This is important when interpreting plots produced.

For this exercise, we will use the regime dataset and include the categorical variables un_continent_name (eg: Asia, North America,…), the regime type (e.g., monarchy, civilian,…) and the year the regime started in, start_year. The estimator to fit unknown coefficients in Aalen’s additive model is located under AalenAdditiveFitter.

from lifelines import AalenAdditiveFitter
from lifelines.datasets import load_dd

data = load_dd()
ctryname cowcode2 politycode un_region_name un_continent_name ehead leaderspellreg democracy regime start_year duration observed
Afghanistan 700 700 Southern Asia Asia Mohammad Zahir Shah Mohammad Zahir Shah.Afghanistan.1946.1952.Monarchy Non-democracy Monarchy 1946 7 1
Afghanistan 700 700 Southern Asia Asia Sardar Mohammad Daoud Sardar Mohammad Daoud.Afghanistan.1953.1962.Civilian Dict Non-democracy Civilian Dict 1953 10 1
Afghanistan 700 700 Southern Asia Asia Mohammad Zahir Shah Mohammad Zahir Shah.Afghanistan.1963.1972.Monarchy Non-democracy Monarchy 1963 10 1
Afghanistan 700 700 Southern Asia Asia Sardar Mohammad Daoud Sardar Mohammad Daoud.Afghanistan.1973.1977.Civilian Dict Non-democracy Civilian Dict 1973 5 0
Afghanistan 700 700 Southern Asia Asia Nur Mohammad Taraki Nur Mohammad Taraki.Afghanistan.1978.1978.Civilian Dict Non-democracy Civilian Dict 1978 1 0

I’m using the lovely library Patsy here to create a design matrix from my original dataframe.

import patsy
X = patsy.dmatrix('un_continent_name + regime + start_year', data, return_type='dataframe')
X = X.rename(columns={'Intercept': 'baseline'})

 'regime[T.Military Dict]',
 'regime[T.Mixed Dem]',
 'regime[T.Parliamentary Dem]',
 'regime[T.Presidential Dem]',

We have also included the coef_penalizer option. During the estimation, a linear regression is computed at each step. Often the regression can be unstable (due to high co-linearity or small sample sizes) – adding a penalizer term controls the stability. I recommend always starting with a small penalizer term – if the estimates still appear to be too unstable, try increasing it.

aaf = AalenAdditiveFitter(coef_penalizer=1.0, fit_intercept=False)

An instance of AalenAdditiveFitter includes a fit() method that performs the inference on the coefficients. This method accepts a pandas DataFrame: each row is an individual and columns are the covariates and two individual columns: a duration column and a boolean event occurred column (where event occurred refers to the event of interest - expulsion from government in this case)

X['T'] = data['duration']
X['E'] = data['observed'], 'T', event_col='E')

After fitting, the instance exposes a cumulative_hazards_ DataFrame containing the estimates of \(\int_0^t b_i(s) \; ds\):

baseline un_continent_name[T.Americas] un_continent_name[T.Asia] un_continent_name[T.Europe] un_continent_name[T.Oceania] regime[T.Military Dict] regime[T.Mixed Dem] regime[T.Monarchy] regime[T.Parliamentary Dem] regime[T.Presidential Dem] start_year
-0.03447 -0.03173 0.06216 0.2058 -0.009559 0.07611 0.08729 -0.1362 0.04885 0.1285 0.000092
0.14278 -0.02496 0.11122 0.2083 -0.079042 0.11704 0.36254 -0.2293 0.17103 0.1238 0.000044
0.30153 -0.07212 0.10929 0.1614 0.063030 0.16553 0.68693 -0.2738 0.33300 0.1499 0.000004
0.37969 0.06853 0.15162 0.2609 0.185569 0.22695 0.95016 -0.2961 0.37351 0.4311 -0.000032
0.36749 0.20201 0.21252 0.2429 0.188740 0.25127 1.15132 -0.3926 0.54952 0.7593 -0.000000

AalenAdditiveFitter also has built in plotting:

aaf.plot(columns=['regime[T.Presidential Dem]', 'baseline', 'un_continent_name[T.Europe]'], iloc=slice(1,15))

Regression is most interesting if we use it on data we have not yet seen, i.e., prediction! We can use what we have learned to predict individual hazard rates, survival functions, and median survival time. The dataset we are using is available up until 2008, so let’s use this data to predict the duration of former Canadian Prime Minister Stephen Harper.

ix = (data['ctryname'] == 'Canada') & (data['start_year'] == 2006)
harper = X.loc[ix]
print("Harper's unique data point:")
Harper's unique data point:
     baseline  un_continent_name[T.Americas]  un_continent_name[T.Asia] ...  start_year  T  E
268       1.0                            1.0                        0.0 ...      2006.0  3  0
ax = plt.subplot(2,1,1)

ax = plt.subplot(2,1,2)


Because of the nature of the model, estimated survival functions of individuals can increase. This is an expected artifact of Aalen’s additive model.

Custom Parametric Regression Models

lifelines has a very general syntax for creating your own parametric regression models. If you are looking to create your own custom models, see docs Custom Regression Models.

Model selection in survival regression

Parametric vs Semi-parametric models

Above, we’ve displayed two semi-parametric models (Cox model and Aalen’s model), and a family of parametric AFT models. Which should you choose? What are the advantages and disadvantages of either? I suggest reading the two following StackExchange answers to get a better idea of what experts think:

  1. In survival analysis, why do we use semi-parametric models (Cox proportional hazards) instead of fully parametric models?
  2. In survival analysis, when should we use fully parametric models over semi-parametric ones?

Model selection based on residuals

The sections Testing the Proportional Hazard Assumptions and Assessing Cox model fit using residuals may be useful for modeling your data better.


Work is being done to extend residual methods to AFT models. Stay tuned.

Model selection based on predictive power

If censoring is present, it’s not appropriate to use a loss function like mean-squared-error or mean-absolute-loss. Instead, one measure is the concordance-index, also known as the c-index. This measure evaluates the accuracy of the ranking of predicted time. It is in fact a generalization of AUC, another common loss function, and is interpreted similarly:

  • 0.5 is the expected result from random predictions,
  • 1.0 is perfect concordance and,
  • 0.0 is perfect anti-concordance (multiply predictions with -1 to get 1.0)

Fitted survival models typically have a concordance index between 0.55 and 0.75 (this may seem bad, but even a perfect model has a lot of noise than can make a high score impossible). In lifelines, a fitted model’s concordance-index is present in the output of print_summary(), but also available under the score_ property. Generally, the measure is implemented in lifelines under lifelines.utils.concordance_index() and accepts the actual times (along with any censored subjects) and the predicted times.

from lifelines import CoxPHFitter
from lifelines.datasets import load_rossi

rossi = load_rossi()

cph = CoxPHFitter(), duration_col="week", event_col="arrest")

# Three ways to view the c-index:
# method one

# method two

# method three
from lifelines.utils import concordance_index
print(concordance_index(rossi['week'], -cph.predict_partial_hazard(rossi), rossi['arrest']))


Remember, the concordance score evaluates the relative rankings of subject’s event times. Thus, it is scale and shift invariant (i.e. you can multiple by a positive constant, or add a constant, and the rankings won’t change). A model maximized for concordance-index does not necessarily give good predicted times, but will give good predicted rankings.

However, there are other, arguably better, methods to measure the fit of a model. Included in print_summary is the log-likelihood, which can be used in an AIC calculation, and the log-likelihood ratio statistic. Generally, I personally loved this article by Frank Harrell, “Statistically Efficient Ways to Quantify Added Predictive Value of New Measurements”.

lifelines has an implementation of k-fold cross validation under lifelines.utils.k_fold_cross_validation(). This function accepts an instance of a regression fitter (either CoxPHFitter of AalenAdditiveFitter), a dataset, plus k (the number of folds to perform, default 5). On each fold, it splits the data into a training set and a testing set fits itself on the training set and evaluates itself on the testing set (using the concordance measure by default).

from lifelines import CoxPHFitter
from lifelines.datasets import load_regression_dataset
from lifelines.utils import k_fold_cross_validation

regression_dataset = load_regression_dataset()
cph = CoxPHFitter()
scores = k_fold_cross_validation(cph, regression_dataset, 'T', event_col='E', k=3)

#[ 0.5896  0.5358  0.5028]
# 0.542
# 0.035

Also, lifelines has wrappers for compatibility with scikit learn for making cross-validation and grid-search even easier.