fix renv and minor bugs

06-sl3.Rmd

@@ -67,15 +67,17 @@ The Super Learner (SL) solves the issue of selecting an algorithm, as it can
 consider many of them - from the simplest parametric regressions to the most
 complex machine learning algorithms (e.g., neural nets, support vector machines,
 etc). Additionally, it is proven to perform as well as possible 
-(as good as the unknown oracle) in large samples, given the learners specified [@vdl2003unified; @vdl2004asymptotic; @dudoit2005asymptotics; @vaart2006oracle]. The SL represents an entirely pre-specified, flexible, and theoretically grounded approach for
-predictive modeling. It has been shown to be adaptive and robust in a variety of
-applications, even in very small samples. Detailed descriptions outlining
-the SL procedure are widely available [@polley2010super; @naimi2018stacked].
-Practical considerations for specifying the SL, including how to specify a rich
-and diverse library of learners, choose a performance metric for the SL, and
-specify a cross-validation (CV) scheme, are described in a pre-print article
-[@rvp2022super]. Here, we focus on introducing `sl3`, the standard `tlverse`
-software package for SL.
+(as good as the unknown oracle) in large samples, given the learners specified
+[@vdl2003unified; @vdl2004asymptotic; @dudoit2005asymptotics; @vaart2006oracle].
+The SL represents an entirely pre-specified, flexible, and theoretically
+grounded approach for predictive modeling. It has been shown to be adaptive and
+robust in a variety of applications, even in very small samples. Detailed
+descriptions outlining the SL procedure are widely available [@polley2010super;
+@naimi2018stacked]. Practical considerations for specifying the SL, including
+how to specify a rich and diverse library of learners, choose a performance
+metric for the SL, and specify a cross-validation (CV) scheme, are described in
+a pre-print article [@rvp2022super]. Here, we focus on introducing `sl3`, the
+standard `tlverse` software package for SL.
 
 <!--
 IM:
@@ -104,12 +106,13 @@ Fitting any SL with `sl3` consists of the following three steps:
 
 #### Running example with WASH Benefits dataset {-}
 
-We will use the WASH Benefits Bangladesh study as an example to guide this 
-overview of `sl3`. In this study, we are interested in predicting the child development outcome, weight-for-height z-score, from covariates/predictors, 
-including socio-economic status variables, gestational age, and maternal 
-features. More information on this dataset is described in the ["Meet 
-the Data"](https://tlverse.org/tlverse-handbook/data.html#wash) chapter of the 
-`tlverse` handbook. 
+We will use the WASH Benefits Bangladesh study as an example to guide this
+overview of `sl3`. In this study, we are interested in predicting the child
+development outcome, weight-for-height z-score, from covariates/predictors,
+including socio-economic status variables, gestational age, and maternal
+features. More information on this dataset is described in the ["Meet the
+Data"](https://tlverse.org/tlverse-handbook/data.html#wash) chapter of the
+`tlverse` handbook.
 
 #### Preliminaries {-}
 
@@ -123,8 +126,8 @@ library(kableExtra)
 library(data.table)
 ```
 
-We will use the `fread` function in the `data.table` R package to load the 
-WASH Benefits example dataset:
+We will use the `fread` function in the `data.table` R package to load the WASH
+Benefits example dataset:
 
 ```{r load-data}
 library(data.table)
@@ -1127,7 +1130,7 @@ Recall the SL object, `sl`, defined in section 2:
 ```{r recall-eSL}
 # in the section 2 we defined Lrnr_sl as
 # sl <- Lrnr_sl$new(learners = stack, metalearner = Lrnr_nnls$new())
-````
+```
 
 `sl` is an eSL since it used NNLS as the meta-learner. We rename `sl` to 
 `eSL_metaNNLS` below to clarify that this is an eSL that uses NNLS as its