-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy path3_packages.Rmd
126 lines (73 loc) · 6.49 KB
/
3_packages.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
---
title: "Packages"
author: "Irfan Kanat"
date: "12/20/2015"
output:
html_document:
df_print: paged
pdf_document: default
---
We have already encountered packages quite a few times so far. In introduction we discussed the packages and the opportunities they provided. In importing we have used them to read in data in various formats. Yet, we have so far avoided a serious discussion of what packages are and how we can use them.
Packages are a collection of functions, documentation, and in some cases data files. These are similar to libraries in other programing languages. **When loaded into memory**, they provide specific functionality, be it analysis, visualization, or reporting.
R being open source enabled a great number of packages serving any function imaginable, making it one of the the most popular tools for analytics. If a feature is missing, you can easily develop your own functions and bundle them into a package for others to use.
In this document we will go over the basics of package management in R.
# Basics
Before getting into the details of package management, let us learn how to obtain and activate a package.
If the package you need is not already installed in your system, you can easily obtain it from CRAN via install.packages() function. Below you can see the command to install the plm package, a set of functions to carry out traditional econometric analysis.
```{r, eval=F}
install.packages("plm") # Notice the quotes around package name
```
Pretty soon the package will be installed. Installing a package does not mean it is ready to use. You also have to activate it. For this purpose you can use the library() command.
```{r}
library(plm) # Notice the lack of quotes around the package name
```
You most probably won't need this, still it is good to know. To de-activate a package, we use the detach() function. The parameters of this command is a bit more involved than the first two.
```{r}
detach("package:plm", unload=TRUE) # Notice the package: prepended to the package name
```
# Details
If you want to get a list of all the packages that are currently loaded in the memory (activated) use the .packages() command.
```{r}
(.packages()) # Notice the parantheses
```
If you want to get a list of all packages installed in the system use the parameter. I am omitting the output since I have about 150 packages installed.
```{r, eval=FALSE}
.packages(all.available = T)
```
# How to find out about packages?
If you want to find packages to carry out the tasks you have at hand the first place to start would be either [Comprehensive R Archive Network (CRAN)](https://cran.r-project.org/web/packages/) or (for a better interface) [R Forge.](https://r-forge.r-project.org/) [R site search](http://search.r-project.org) (also available with command RSiteSearch()) provides a great search engine for documentation. [R seek](http://rseek.org/) is a great search engine dedicated to scanning R related sources. Of course you can always use google as well but the letter R is a bit ambiguous in some cases.
R frequently offers a number of packages that serves similar purposes (lme4 and nlme for mixed models for example). There are subtle differences between the packages and picking the right tool for the task can save you a lot of time. I often find myself using documentation from cran pages of the packages before I settle for the one I want to use. I frequently spend more time browsing documentation, than on actual analysis. A careful reading of the documentation will allow you to better understand the options available to you.
CRAN is not only a repository for the code, but also for the documentation. Here is the [plm package](https://cran.r-project.org/web/packages/plm/) page for example. Beyond reference manuals and other documentation, R packages often come with vignettes and tutorials. These are often like quick start guides to analysis.
If you want to search a certain word in installed packages' documentation, you can always use ?? or help.search()
```{r}
??mixed
help.search("mixed model")
```
## Commonly Used Packages
### Data Manipulation
#### data.table
This package is a replacement of the data.frame data structure. It is not a drop-in replacement and is not fully compatible with all data.frame functionality. So you need to be careful when you use it.
It replaces data.frame with a data structure that is more efficient. Meaning it is faster when accessing and writing into data.tables.
It also comes with additional functionality that makes subsetting/aggregating a breeze.
#### dplyr
This is a package that adds additional functions to efficiently manipulate data (subsetting/aggregate, etc.). It still uses data.frames so you do not have compatibility issues in your code.
Another benefit is, it can directly work with data stored in databases. So you can select, filter data from a database and use it in R.
### Statistics
R base package (that comes loaded) includes your basic statistics such as multiple regressions and Generalized Linear Models.
#### Panel Data Econometrics
Your traditional fixed effects / random effects econometric models are covered in plm package.
#### Hierarchical/Mixed Effects Models
Two packages, nlme and lme4 handle this bit. The development on nlme has slowed down a bit, yet it incorporates functionality to model autoregressive correlation structures, weighting to take care of heteroskedasticity and so on. Both packages have unique strengths, you need to read the documentation to see which one fits you best.
### Machine Learning
caret package provides a nice set of functions for almost all predictive analytics needs. It is discussed in deeper detail in Example Analysis.
#### Classifiers (KNN, LVQ)
K Nearest Neighbor matching and Learning Vector Quantization and more are available through class package.
#### Support Vector Machines
kernlab, e1071 packages are two among many packages that provide this functionality.
#### Clustering
Base package has functionality to enable basic clustering techniques like K means (kmeans()), or hierarchical (hclust()) clustering.
For model based clustering there is the mclust package.
#### Neural Networks
neuralnet package as well as others provide NN functionality.
------
 How I Learned to Stop Worrying and Love the R Console by [Irfan E Kanat](http://irfankanat.com) is licensed under a [Creative Commons Attribution 4.0 International License](http://creativecommons.org/licenses/by/4.0/). Based on a work at [http://github.com/iekanat/rworkshop](http://github.com/iekanat/rworkshop).