top10.Rmd

---
title: "Top 10 Data Mining Algorithms"
author: "weiya"
date: "April 3, 2019"
output: html_document
---

本文参考 [HackerBits's Top 10 data mining algorithms in plain R](https://hackerbits.com/data/top-10-data-mining-algorithms-in-plain-r/)。

## C5.0 (前身是 C4.5)

算法细节详见 [9.2 基于树的方法(CART)](https://esl.hohoweiya.xyz/09-Additive-Models-Trees-and-Related-Methods/9.2-Tree-Based-Methods/index.html)。

```{r}
library(C50)
library(printr)
# divide into traing data and test data
set.seed(1234)
train.idx = sample(1:nrow(iris), 100)
iris.train = iris[train.idx, ]
iris.test = iris[-train.idx, ]
# train C5.0
model = C5.0(Species ~ ., data = iris.train)
# predict
results = predict(object = model, newdata = iris.test, type = "class")
# confusion matrix
table(results, iris.test$Species)
```

## K-means

算法细节详见 [13.2 原型方法](https://esl.hohoweiya.xyz/13-Prototype-Methods-and-Nearest-Neighbors/13.2-Prototype-Methods/index.html)。

```{r}
# run kmeans
model = kmeans(x = subset(iris, select = -Species), centers = 3)
# check results
table(model$cluster, iris$Species)
```

## Support Vector Machine

算法细节详见 [12.2 支持向量分类器](https://esl.hohoweiya.xyz/12-Support-Vector-Machines-and-Flexible-Discriminants/12.2-The-Support-Vector-Classifier/index.html)。

```{r}
library(e1071)
model = svm(Species ~ ., data = iris.train)
results = predict(object = model, newdata = iris.test, type = "class")
table(results, iris.test$Species)
```

## Apriori

算法细节详见 [14.2 关联规则](https://esl.hohoweiya.xyz/14-Unsupervised-Learning/14.2-Association-Rules/index.html)。

```{r, message=FALSE}
library(arules)
data("Adult")
rules = apriori(Adult,
                parameter = list(support = 0.4, confidence = 0.7),
                appearance = list(rhs = c("race=White", "sex=Male"), default = "lhs"))
# results
rules.sorted = sort(rules, by = "lift")
top5.rules = head(rules.sorted, 5)
as(top5.rules, "data.frame")
```

## EM

算法细节详见 [8.5 EM 算法](https://esl.hohoweiya.xyz/08-Model-Inference-and-Averaging/8.5-The-EM-Algorithm/index.html)。

```{r}
library(mclust)
model = Mclust(subset(iris, select = -Species))
table(model$classification, iris$Species)
```

## Pagerank

算法细节详见 [14.10 谷歌的 PageRank 算法](https://esl.hohoweiya.xyz/14-Unsupervised-Learning/14.10-The-Google-PageRank-Algorithm/index.html)

```{r, message = FALSE}
library(igraph)
library(dplyr)
# generate a random directed graph
set.seed(111)
g = random.graph.game(n = 10, p.or.m = 1/4, directed = TRUE)
plot(g)
# calculate the pagerank for each object
pr = page.rank(g)$vector
# view results
df = data.frame(Object = 1:10, PageRank = pr)
arrange(df, desc(PageRank))
```

## AdaBoost

算法细节详见 [10.1 boosting 方法](https://esl.hohoweiya.xyz/10-Boosting-and-Additive-Trees/10.1-Boosting-Methods/index.html)。

```{r, message=FALSE}
library(adabag)
model = boosting(Species ~ ., data = iris.train)
results = predict(object = model, newdata = iris.test, type = "class")
results$confusion
```

## kNN

算法细节详见 [k 最近邻分类器](https://esl.hohoweiya.xyz/13-Prototype-Methods-and-Nearest-Neighbors/13.3-k-Nearest-Neighbor-Classifiers/index.html)。

```{r, message=FALSE}
library(class)
results = knn(train = subset(iris.train, select = -Species),
              test = subset(iris.test, select = -Species),
              cl = iris.train$Species)
table(results, iris.test$Species)
```

## Naive Bayes

算法细节详见 [6.6 核密度估计和分类](https://esl.hohoweiya.xyz/06-Kernel-Smoothing-Methods/6.6-Kernel-Density-Estimation-and-Classification/index.html)。

```{r}
model = naiveBayes(x = subset(iris.train, select = -Species), 
                   y = iris.train$Species)
results = predict(object = model, newdata = iris.test, type = "class")
table(results, iris.test$Species)
```

## CART

算法细节详见 [9.2 基于树的方法(CART)](https://esl.hohoweiya.xyz/09-Additive-Models-Trees-and-Related-Methods/9.2-Tree-Based-Methods/index.html)。

```{r}
library(rpart)
model = rpart(Species ~ ., data = iris.train)
results = predict(object = model, newdata = iris.test, type = "class")
table(results, iris.test$Species)
```