Documentation fixes

JuliaStats · Jan 24, 2019 · 9547b10 · 9547b10
1 parent c519a8e
commit 9547b10
Show file tree

Hide file tree

Showing 9 changed files with 47 additions and 33 deletions.
diff --git a/docs/Project.toml b/docs/Project.toml
@@ -1,5 +1,6 @@
 [deps]
 Documenter = "e30172f5-a6a5-5a46-863b-614d45cd2de4"
+StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
 Survival = "8a913413-2070-5976-9d4c-2b364fdc2f7f"
 
 [compat]

diff --git a/docs/make.jl b/docs/make.jl
@@ -1,4 +1,4 @@
-using Survival, Documenter
+using Survival, Documenter, StatsBase
 
 makedocs(
     modules = [Survival],

diff --git a/docs/src/cox.md b/docs/src/cox.md
@@ -1,12 +1,17 @@
 # Cox Proportional Hazards Model
 
-The [Cox proportional hazards model](https://en.wikipedia.org/wiki/Proportional_hazards_model) is a semiparametric regression used to fit survival models without knowing the distribution. It is based on the assumption that different covariates affect the hazard function multiplicatively:
+The [Cox proportional hazards model](https://en.wikipedia.org/wiki/Proportional_hazards_model)
+is a semiparametric regression model used to fit survival models without knowing the
+distribution. It is based on the assumption that covariates affect the hazard function
+multiplicatively. That is,
 
 ```math
-\lambda(t|X_i) = \lambda_0(t)\exp(X_i\cdot\beta)
+\lambda(t | X_i) = \lambda_0(t) \exp(X_i \cdot \beta)
 ```
 
-where ``\lambda(t|X_i)`` is the estimated hazard for sample ``i``, ``\lambda_0`` is the baseline hazard, ``X_i`` the vector of covariates for sample ``i`` and ``\beta`` are the coefficients of the regression.
+where ``\lambda(t|X_i)`` is the estimated hazard for sample ``i``, ``\lambda_0`` is the
+baseline hazard, ``X_i`` is the vector of covariates for sample ``i``, and ``\beta`` is
+the vector of coefficients in the model.
 
 ## API
 
@@ -16,4 +21,5 @@ StatsBase.fit(::Type{CoxModel}, M::AbstractMatrix, y::AbstractVector; kwargs...)
 
 ## References
 
-* Cox, D. R. (1972). *Regression models and life tables (with discussion)*. Journal of the Royal Statistical Society, Series B, 34:187–220.
+* Cox, D. R. (1972). *Regression models and life tables (with discussion)*.
+  Journal of the Royal Statistical Society, Series B, 34:187–220.
diff --git a/docs/src/index.md b/docs/src/index.md
@@ -1,16 +1,17 @@
+```@meta
+DocTestSetup = :(using Survival, StatsBase)
+CurrentModule = Survival
+```
+
 # Survival.jl
 
 This package provides types and methods for performing
 [survival analysis](https://en.wikipedia.org/wiki/Survival_analysis) in Julia.
 
 ## Installation
 
-The package is not yet registered in Julia's package registry, and so it must
-be installed using
-
-```julia
-Pkg.clone("https://github.com/ararslan/Survival.jl.git")
-```
+The package is not yet registered in Julia's General package registry, and so it must
+be installed using `Pkg.add("https://github.com/ararslan/Survival.jl")`.
 
 ## Contents
 

diff --git a/docs/src/km.md b/docs/src/km.md
@@ -24,10 +24,8 @@ using Greenwood's formula:
 
 ```@docs
 Survival.KaplanMeier
-StatsBase.fit(::Type{S},
-              times::AbstractVector{T},
-              status::AbstractVector{<:Integer}) where {S<:NonparametricEstimator, T}
-StatsBase.confint(km::KaplanMeier, α::Float64=0.05)
+StatsBase.fit(::Type{KaplanMeier}, ::Any, ::Any)
+StatsBase.confint(::KaplanMeier, ::Float64)
 ```
 
 ## References

diff --git a/docs/src/na.md b/docs/src/na.md
@@ -13,8 +13,8 @@ where ``d_i`` is the number of observed events at time ``t_i`` and ``n_i`` is th
 number of subjects at risk for the event just before time ``t_i``.
 
 The pointwise standard error of the log of the survivor function can be computed
-directly as the standard error or a Bernoulli random variable with `d_i` successes
-from `n_i` samples:
+directly as the standard error or a Bernoulli random variable with ``d_i`` successes
+from ``n_i`` samples:
 
 ```math
 \text{SE}(\hat{H}(t)) = \sqrt{\sum_{i: t_i < t} \frac{d_i(n_i-d_i)}{n_i^3}}
@@ -24,13 +24,11 @@ from `n_i` samples:
 
 ```@docs
 Survival.NelsonAalen
-StatsBase.fit(::Type{S},
-              times::AbstractVector{T},
-              status::AbstractVector{<:Integer}) where {S<:NonparametricEstimator, T}
-StatsBase.confint(na::NelsonAalen, α::Float64=0.05)
+StatsBase.fit(::Type{NelsonAalen}, ::Any, ::Any)
+StatsBase.confint(::NelsonAalen, ::Float64)
 ```
 
 ## References
 
 * Nelson, W. (1969). *Hazard plotting for incomplete failure data*.
-  Journal of Quality Technology 1, 27–52.
+  Journal of Quality Technology 1, 27–52.
diff --git a/src/estimator.jl b/src/estimator.jl
@@ -55,17 +55,9 @@ function _estimator(::Type{S}, tte::AbstractVector{T}, status::BitVector) where
     return S{T}(times, nevents, ncensor, natrisk, estimator, stderr)
 end
 
-"""
-    fit(::Type{S}, times, status) where S<:NonparametricEstimator
-
-Given a vector of times to events and a corresponding vector of indicators that
-dictate whether each time is an observed event or is right censored, compute the
-model of type `S`. Return an object of type `S`: [`KaplanMeier`](@ref) and 
-[`NelsonAalen`](@ref) are supported so far.
-"""
 function StatsBase.fit(::Type{S},
                        times::AbstractVector{T},
-                       status::AbstractVector{<:Integer}) where {S<:NonparametricEstimator, T}
+                       status::AbstractVector{<:Integer}) where {S<:NonparametricEstimator,T}
     nobs = length(times)
     if length(status) != nobs
         throw(DimensionMismatch("there must be as many event statuses as times"))
@@ -88,4 +80,4 @@ function StatsBase.fit(::Type{S}, ets::AbstractVector{<:EventTime}) where S<:Non
     # make this method the One True Method™ so that folks are encouraged to
     # use EventTimes instead of raw values.
     return fit(S, map(t->t.time, x), BitVector(map(t->t.status, x)))
-end
+end
diff --git a/src/kaplanmeier.jl b/src/kaplanmeier.jl
@@ -44,3 +44,12 @@ function StatsBase.confint(km::KaplanMeier, α::Float64=0.05)
         exp(-exp(l - a)), exp(-exp(l + a))
     end
 end
+
+"""
+    fit(KaplanMeier, times, status) -> KaplanMeier
+
+Given a vector of times to events and a corresponding vector of indicators that
+dictate whether each time is an observed event or is right censored, compute the
+Kaplan-Meier estimate of the survivor function.
+"""
+StatsBase.fit(::Type{KaplanMeier}, times, status)
diff --git a/src/nelsonaalen.jl b/src/nelsonaalen.jl
@@ -40,4 +40,13 @@ function StatsBase.confint(na::NelsonAalen, α::Float64=0.05)
     return map(na.chaz, na.stderr) do srv, se
         srv - q * se, srv + q * se
     end
-end
+end
+
+"""
+    fit(NelsonAalen, times, status) -> NelsonAalen
+
+Given a vector of times to events and a corresponding vector of indicators that
+dictate whether each time is an observed event or is right censored, compute the
+Nelson-Aalen estimate of the cumulative hazard rate function.
+"""
+StatsBase.fit(::Type{NelsonAalen}, times, status)