首页 > 代码库 > qplot()函数的详细用法

qplot()函数的详细用法

qplot()函数的详细用法:

library(ggplot2)

# 测试数据集,ggplot2内置的钻石数据
qplot(carat, price, data = http://www.mamicode.com/diamonds)
dsmall <- diamonds[sample(nrow(diamonds), 100), ] #对diamonds数据集进行抽样

#1. 按color,size,shape的基本分类可视化

#1.1 简单的散点图(利用color分类,不同颜色的钻石由不同颜色的点代表)
qplot(carat, price, data = http://www.mamicode.com/dsmall, colour = color)

技术分享
 



#1.2. 简单的散点图(利用shape分类,不同的切割方式由不同形状的点代表)
qplot(carat, price, data = http://www.mamicode.com/dsmall, shape = cut)

技术分享
 



#2. 绘制不同类型的图表:geom参数

qplot(x,y,data=http://www.mamicode.com/data,geom="")中的geom=""用来控制输出的图形类型
I. 两变量图
(1) geom="points",默认参数,绘制散点图(x,y)
(2) geom="smooth" 绘制平滑曲线(基于loess, gam, lm ,rlm,glm)
(3) geom="boxplot" 绘制箱线图 ,当x为属性变量(factor),y为数值变量时

II.单变量图
(4) geom="histogram",直方图
(5) geom="density",核密度估计图
(6) geom="bar",条形图barchart

III.时间序列
(7) geom="line",折线图,可用于时间序列(当x=date)
(8) geom="path",路径图(参见后文)


# 2.1 同时绘制散点图+平滑直线
qplot(carat, price, data = http://www.mamicode.com/dsmall, geom=c("point","smooth"))

技术分享
 



#参数调整:method=""等
#(a). method = "loess", 默认平滑算法, 通过span=调整窗宽, span=0(波动) 到 span=1(光滑)
qplot(carat, price, data = http://www.mamicode.com/dsmall, geom = c("point", "smooth"),
      method = "loess",span=0.2)

技术分享
 



# (b). method = "gam": GAM 在大数据时比loess高效,需要载入 mgcv 包
library(mgcv)
qplot(carat, price, data = http://www.mamicode.com/dsmall, geom = c("point", "smooth"),
      method="gam", formula = y ~ s(x))

# (c). method="lm", 线性平滑
qplot(carat, price, data = http://www.mamicode.com/dsmall, geom = c("point", "smooth"),
     method = "lm")

# method="lm",formula = y ~ ns(x, 3),三次自然样条,需要载入splines包
library(splines)
qplot(carat, price, data = http://www.mamicode.com/dsmall, geom = c("point", "smooth"),
      method = "lm", formula = y ~ ns(x, 3))

# method = "rlm", robust linear model, 受异常值影响小,需要载入MASS包
library(MASS)
qplot(carat, price, data = http://www.mamicode.com/dsmall, geom = c("point", "smooth"),
      method = "rlm")


# 2.2:x为属性变量,y为连续变量,绘制boxplot
qplot(color, price/carat, data=http://www.mamicode.com/diamonds,geom="boxplot")

技术分享
 



# 2.3:单变量,直方图
qplot(carat, data = http://www.mamicode.com/diamonds, geom ="histogram")

技术分享
 



#2.4: 单变量,核密度估计图
qplot(carat, data = http://www.mamicode.com/diamonds, geom ="density")

技术分享
 



# 按不同颜色绘制的density图
qplot(carat, data = http://www.mamicode.com/diamonds, geom ="density",colour=color)

技术分享
 



# 2.5 条形图(柱状图)
#计数,求count(color)
qplot(color, data = http://www.mamicode.com/diamonds, geom ="bar")

#加权,对每个求sum(carat),类似于excel里的数据透视图,按不同的color计算carat的总和
qplot(color, data = http://www.mamicode.com/diamonds, geom ="bar", weight = carat)

技术分享
 




#2.6. Time-series
qplot(date, unemploy / pop, data = http://www.mamicode.com/economics, geom ="line")

技术分享
 




#2.7. Path plot
#如果要查看失业率(unemploy / pop)与平均失业时间(uempmed)之间的关系,一个方法是利用散点图,但是这样做就会导致无法观察到随时间变化的趋势了,path plot利用颜色深浅来代表年份,随着颜色从浅蓝变成深蓝,可以观察到失业率与失业时间的关系的变化趋势。

#具体实现:先自定义函数year(),将字符串格式的时间转化为年
year <- function(x) as.POSIXlt(x)$year + 1900

#画出path plot,颜色按年份由浅到深
qplot(unemploy / pop, uempmed, data = http://www.mamicode.com/economics,
      geom = "path", colour = year(date))

技术分享

qplot()函数的详细用法