Share
Embed
Transcript
36 slides
-
F3S Workshop Try R A Statistical Data Analysis Tool Takahiro
Sumiya Information Media Center, Hiroshima University -
‣ R is a language and environment for statistic computing
and graphics https://www.r-project.org/about.html -
What you want to do Algorithm 1: int main(int argc,char
2: int i=0; 3: char c; 4: while(i==0){ 5: c=getchar(); Program Do it! language environment -
Agenda 1. Preparation (Installing Studio) 2. Invoking command and find
the result 3. Importing data from Excel 4. Summarizing data 5. Visualizing data shape 6. Try a few statistical methods 4 -
Rͷ४උ ‣ R ✓ Engine of R ✓ Include minimum
environment ‣ RStudio ✓ More convenience environment ✓ For your daily use ✓ It requires Engine of R 6 ʴ -
RͷΠϯετʔϧ 7 https://www.r-project.org/about.html Select mirror site in Japanese CRAN page
Download R for (Mac) OS X → R-3.5.1.pkg Download R for Windows → Install R for the first time ɹɹɹɹɹɹɹɹɹɹˠ Download R 3.5.1 for Windows -
RStudio ͷΠϯετʔϧ 8 https://www.rstudio.com at the bottom RStudio 1.1.463 -
Windows Vista/7/8/10 RStudio 1.1.463 - Mac OS X 10.6+ (64-bit) -
R is a calculator ‣ Try to input mathematical formula
> 1+1 > 100/3 > 100/3*3 ‣ You can use parenthesis > (3+5)*7 > 3+5*7 # check the result 12 -
Save R commands to a File ‣ You can bind
several files in "Project" ✓ Command file (script file) ✓ Data file ✓ Visualization ‣ File → New Project ɹɹ → New Directory → New Project 13 -
Save R commands to a File ‣ File→New File→R Script
14 Type and ctl-enter Results -
R is a high level calculator ‣ Power > 2^16
‣ Functions > sin(pi/2) # > exp(1) # > factorial(10) # > choose(5,2) # 15 sin ⇡ 2 10! 5C2 e1 -
Try graph > plot(c(5,5,4,3,3,4,1,1,1)) > x=c(5,4,3,3,1,4,1,1,1) # variable definition >
plot(x) # simple! > plot(x,type="b") # What is this data? > plot(x,type="b",ylim=c(6,1)) > yr=2010:2018 > plot(yr,x,type="b",ylim=c(6,1)) 16 -
In R, variable (object) is vector > x # Just
type name of variable > x+10 # Check the result > x+c(10,100) > x[1] # The first value of vector > x[c(1,3,5)] # the 1st, 3rd, 5th values 17 -
Close and re-open it ‣ Save script file ‣ Quit
RStudio ‣ Check the script file in folder ‣ Re-open it by double-clicking the project file 18 -
Import Excel data with CSV ‣ Download "carp-e.xlsx" from Bb9
‣ Open it with Excel and check it ‣ Save as "CSV (UTF-8)" → carp.csv ‣ Save "CSV" on Windows 20 -
"Data frame" > read.csv("carp.csv") # just print > c=read.csv("carp.csv") #
save in "c" > c # print the content ‣ You can bind several vectors with name in "data frame" > c$height # vector with name "height" > c$height[3] # third value of the vector > mean(c$height) # average of "height" 21 -
"Data frame" (cont.) > c$BMI=c$weight/(c$weight/100)^2 #Calcurate Body Mass Index >
c[c$BMI>30,] # Players whose BMI is lager than 30 22 BMI = Weight(kg) Height(m) 2 -
Advance 1: You can read Excel file directly ‣ install.packages("readxl")
‣ library(readxl) ‣ carp=read_excel("carp.xlsx",1) ‣ dragons=read_excel("carp.xlsx",2) 23 -
At first, try summary() ✓ summary(c) ‣ Numerical data ✓
Min(Minimum value),1st Qu.(the 1st quarter), Median, Mean, 3rd Qu. (the 3rd quarter), Max(Maximum value) ‣ Categorical data ✓ Frequency 25 -
Summary of data group > # summary of players whose
BMI is lager than 26 > summary(c[c$BMI>26,]) > # summary of height by position > tapply(c$height,c$position,summary) 26 -
On Mac, you need to set font to use Japanese
in graphics > # ヒラギノ角ゴシックをW3使うように指定 > par(family = "HiraKakuProN-W3") 28 -
Histgram, scatter plot, box-wisker plot > hist(c$height) # histgram >
barplot(table(c$position)) > plot(c$height,c$weight) # scatter plot > plot(c) # scatterplot matrix > plot(c[,c(4,5,6,9)]) > boxplot(c$height) # box-wicker plot > boxplot(c$height[c$position=="pitcher"], c$height[c$position!="pitcher"], names=c("pitcher","other"),ylab="height") 29 -
Advance 2: Package for psychology > install.packages("psych") > library(psych) >
pairs.panels(c[,c(4,5,6,9)]) 30 -
Linear regression > plot(c$weight ~ c$height) > result=lm(c$weight ~ c$height)
> abline(result) > summary(result) 34 -
Regarding baseball player, is lefty ratio significantly large? Put the
usual ratio to 0.1, try the binomial test. > summary(c$throwing) > binom.test(13,13+56,0.1) > cp=c[c$position=="pitcher",] > summary(cp$throwing) > binom.test(11,11+23,0.1) 35 -
The advantage of R ‣ Designed for statistic calculation ‣
Beautiful graphics ‣ Operations are recorded as script (=text), so it will be re-played easily ‣ New statistical methods are going to be implemented on R ‣ It's free! Open source! 36