- Keep your data in text files
- Read data from a text file
- Modify your data frame to simplify the analysis
- select rows and columns
- add or modify columns
- Calculate summary statistics and models
- Draw nice plots to show your results
- Write text explaining the results
December 13, 2018
Summary of the course
Summary
What will be your grade?
It is hard to make predictions, especially about the future
We can try to predict based in the past
We know the midterm grades, the attendance to classes and quizzes, and the survey you filled at the beginning of the semester
First data: official list from Aksis
head(aksis)
Numara Ad Soyad before 1 0405140001 GÜLSELİ KIRGIL TRUE 2 0405140013 TUĞÇE CAN TRUE 3 0405140023 AYŞE DALLI FALSE 4 0405140028 BÜŞRA KAYIN TRUE 5 0405140051 FİKRİYE ÖZKAYA TRUE 6 0405140052 İLYAS BERKTAŞ FALSE
Second data: Class attendance
head(att)
Numara att12 att21 1 0405140001 0 0 2 0405140013 3 9 3 0405140023 12 21 4 0405140028 1 3 5 0405140051 0 0 6 0405140052 9 15
Combining two data frames
The two data frames have one column in common
cmb1 <- merge(aksis, att, by="Numara") head(cmb1)
Numara Ad Soyad before att12 att21 1 0405140001 GÜLSELİ KIRGIL TRUE 0 0 2 0405140013 TUĞÇE CAN TRUE 3 9 3 0405140023 AYŞE DALLI FALSE 12 21 4 0405140028 BÜŞRA KAYIN TRUE 1 3 5 0405140051 FİKRİYE ÖZKAYA TRUE 0 0 6 0405140052 İLYAS BERKTAŞ FALSE 9 15
Combine with Quizzes delivery
head(quiz)
Numara nquiz 1 0405140001 0 2 0405140013 0 3 0405140023 0 4 0405140028 1 5 0405140051 0 6 0405140052 0
cmb1 <- merge(cmb1, quiz, by="Numara")
Combine with Midterm grades
head(midterm)
Numara grade 1 0405140013 64 2 0405140023 14 3 0405140028 23 4 0405140051 0 5 0405140052 0 6 0405140070 52
cmb1 <- merge(cmb1, midterm, by="Numara")
Combine with Survey
head(sv, 1)
Numara english believe_professor
1 0405110024 I can speak fluently mostly disagree
believe_internet lazy learner
1 mostly disagree mostly disagree mostly disagree
nocalendar believe_journal chess
1 mostly disagree mostly disagree mostly disagree
improviser hardwork time_mgmt
1 mostly disagree mostly disagree mostly disagree
memory genius curiosity
1 mostly disagree mostly disagree mostly disagree
must_be_correct hate_surveys disciplined
1 mostly disagree mostly disagree mostly disagree
need_order all_in_books like_puzzles
1 mostly disagree mostly disagree mostly disagree
books week_study cmb_study scientist_any_other
1 9 6 1 mostly disagree
scientist_robots scientist_not_creative
1 mostly disagree mostly disagree
scientist_like_artists
1 mostly disagree
cmb1 <- merge(cmb1, sv, by="Numara", all.x=TRUE)
Summary
dim(cmb1)
[1] 63 35
summary(cmb1[c(4,5,7)])
before att12 nquiz
Mode :logical Min. : 0.000 Min. :0.000
FALSE:55 1st Qu.: 8.000 1st Qu.:1.000
TRUE :8 Median :10.000 Median :2.000
Mean : 9.111 Mean :2.413
3rd Qu.:11.000 3rd Qu.:4.000
Max. :12.000 Max. :6.000
Simple plots: vectors
plot(cmb1$grade, pch=19)
not very useful
Histogram
hist(cmb1$grade, col="grey")
Plotting factors
plot(cmb1$english)
Changing margins and orientation
par(mar=c(18, 3, 1,1)) plot(cmb1$english, las=2)
Plotting sideways
par(mar=c(3, 18, 1, 1)) barplot(table(cmb1$english), horiz=TRUE, las=2)
Plotting several
par(mar=c(18, 3, 1,1), mfrow=c(2,3))
Analysis
Grade v/s attendance
plot(grade~att12, data=cmb1, pch=19)
Some people did the course before
plot(grade~att12, data=cmb1, pch=19,
col=ifelse(before,"red","black"))
People with mandatory attendance
plot(grade~att12, data=cmb1, pch=19, subset=!before)
People with mandatory attendance
boxplot(grade~att12, data=cmb1, col="grey", subset=!before)
Grade v/s number of quizzes
plot(grade~nquiz, data=cmb1, pch=19, subset=!before)
Grade v/s number of quizzes
boxplot(grade~nquiz, data=cmb1, subset=!before, col="grey")
Combining attendance and quizzes
plot(grade~att12, data=cmb1, subset=!before,
pch=as.character(nquiz))
Grades v/s attendance for 0 quizzes
plot(grade~att12, data=cmb1, subset=!before & nquiz==0,
xlim=c(5,12), ylim=c(0,100), pch=19)
Grades v/s attendance for 1 quizzes
plot(grade~att12, data=cmb1, subset=!before & nquiz==1,
xlim=c(5,12), ylim=c(0,100), pch=19)
Grades v/s attendance for 2 quizzes
plot(grade~att12, data=cmb1, subset=!before & nquiz==2,
xlim=c(5,12), ylim=c(0,100), pch=19)
Grades v/s attendance for 3 quizzes
plot(grade~att12, data=cmb1, subset=!before & nquiz==3,
xlim=c(5,12), ylim=c(0,100), pch=19)
Grades v/s attendance for 4 quizzes
plot(grade~att12, data=cmb1, subset=!before & nquiz==4,
xlim=c(5,12), ylim=c(0,100), pch=19)
Grades v/s attendance for 5 quizzes
plot(grade~att12, data=cmb1, subset=!before & nquiz==5,
xlim=c(5,12), ylim=c(0,100), pch=19)
Grades v/s attendance for 6 quizzes
plot(grade~att12, data=cmb1, subset=!before & nquiz==6,
xlim=c(5,12), ylim=c(0,100), pch=19)
Combining factors and numeric
Is this a language problem?
par(mar=c(3,10,1,1))
plot(grade ~ english, cmb1, subset=!before, horizontal=TRUE,
col="grey", las=2, xlab="")
Is better the second time?
boxplot(grade ~ before, cmb1, col="grey")
Lazy v/s hard work
plot(lazy ~ hardwork, cmb1, subset=!before)
Lazy v/s grade
plot(lazy ~ grade, cmb1, subset=!before)
Lazy v/s grade
par(mar=c(10,4,1,1)) plot(grade ~ lazy, cmb1, subset=!before,las=2, xlab="")
Hard work v/s grade
plot(hardwork ~ grade, cmb1, subset=!before)
Hard work v/s grade
par(mar=c(10,4,1,1)) plot(grade ~ hardwork, cmb1, subset=!before,las=2, xlab="")
Linear models
Grade v/s language
coef(lm(grade~english+0, data=cmb1, subset=!before))
englishİngilizce bilmiyorum
32.50000
englishI can read and understand technical papers
61.90323
englishI can understand movies without subtitles
55.63636
englishI can speak fluently
36.00000
englishEnglish is my native language
32.00000
Grade v/s laziness
coef(lm(grade~lazy+0, data=cmb1, subset=!before))
lazycompletely disagree lazymostly disagree
57.500 50.125
lazyunsure lazymostly agree
62.350 55.000
lazycompletely agree
31.800
Grade v/s nquiz + attendance
coef(lm(grade~nquiz + att12, data=cmb1, subset=!before))
(Intercept) nquiz att12 -1.898583 9.641798 2.953209
Grade v/s nquiz * attendance
coef(lm(grade~nquiz*att12, data=cmb1, subset=!before))
(Intercept) nquiz att12 nquiz:att12 -6.1246308 13.2493943 3.3712794 -0.3307277
Homework
Download the data to your computer
To read it into the R session, use the command load("cmb.Rdata")
Try different combinations of variables to explain which variables affect the exam grade