Veri tabanının oluşturulması
library(data.table)
library(ggplot2)
library(mice)
##
## Attaching package: 'mice'
## The following object is masked from 'package:stats':
##
## filter
## The following objects are masked from 'package:base':
##
## cbind, rbind
library(plm)
##
## Attaching package: 'plm'
## The following object is masked from 'package:data.table':
##
## between
library(stargazer)
##
## Please cite as:
## Hlavac, Marek (2018). stargazer: Well-Formatted Regression and Summary Statistics Tables.
## R package version 5.2.2. https://CRAN.R-project.org/package=stargazer
- Firma veri seti hazırlanması
nf <- 1000
nt <- 20
ns <- 10
nobs <- nf*nt
kelas <- 0.1
lelas <- 0.2
melas <- 0.75
id <- rep(c(1:nf), each = nt)
year <- rep(c(1:nt), times = nf)
sect <- rep(c(1:ns), each = nobs / ns)
kg <- rnorm(nobs, 0.01, 0.02)
lg <- (kg/10) + rnorm(nobs, 0.01, 0.01)
mg <- (lg/10) + rnorm(nobs, 0.005, 0.01)
tc <- rnorm(nobs, 0.01, 0.01)
fs <- max(0.1, rep(rnorm(nf, 2, 0.5), each = nt))
re <- max(0.1, rep(rnorm(nf, 0, 0.5), each = nt))
ee <- rnorm(nobs, 0, 0.1)
fdata <- data.table(id, year, sect, kg, lg, mg, tc, fs, re, ee)
fdata[, lk := fs*(1 + cumsum(kg)), by = id]
fdata[, ll := fs*(1 + cumsum(lg)), by = id]
fdata[, lm := fs*(1 + cumsum(mg)), by = id]
fdata[, lA := fs*(1 + cumsum(tc)), by = id]
fdata[, fe := max(0.1, rep(rnorm(nf, 0, 0.5)) + mean(lk)/5 + mean(ll)/5), by = id]
fdata[, lq := (lA + kelas*lk + lelas*ll + melas*lm) + ee]
fdata[, lqre := re + (lA + kelas*lk + lelas*ll + melas*lm) + ee]
fdata[, lqfe := fe + (lA + kelas*lk + lelas*ll + melas*lm) + ee]
- Bazı kayıp değerler yaratalım
- % 2.5 oranında tamamen rastsal kayıp değerler
fdata[0.025 > runif(nf*nt), lk := NA]
fdata[0.025 > runif(nf*nt), ll := NA]
fdata[0.025 > runif(nf*nt), lm := NA]
- Firmaların %2.5’i için kayıp değerler yaratalım
fdata[id %in% sample(1:nf, floor(0.025*nf)), lk := NA]
fdata[id %in% sample(1:nf, floor(0.025*nf)), ll := NA]
fdata[id %in% sample(1:nf, floor(0.025*nf)), lm := NA]
Veri setini kontrolü
head(fdata)
## id year sect kg lg mg tc fs
## 1: 1 1 1 0.003436922 0.018036581 0.015180031 -0.011610822 3.664568
## 2: 1 2 1 0.052387616 0.018382570 -0.009806749 0.021432255 3.664568
## 3: 1 3 1 0.020034352 0.001411454 0.002170987 0.011186517 3.664568
## 4: 1 4 1 0.021445091 0.019488009 0.005137284 0.007264164 3.664568
## 5: 1 5 1 -0.009941275 0.015467835 0.014920020 0.012481722 3.664568
## 6: 1 6 1 -0.001466867 0.018948229 0.014751797 0.005541357 3.664568
## re ee lk ll lm lA fe lq
## 1: 1.759557 0.188908390 3.677162 3.730664 3.720196 3.622019 3.137542 7.714923
## 2: 1.759557 -0.052586450 3.869140 3.798028 3.684258 3.700559 3.137542 7.557686
## 3: 1.759557 -0.009281759 3.942558 3.803200 3.692214 3.741553 3.137542 7.656327
## 4: 1.759557 0.267214680 4.021145 NA 3.711040 3.768173 3.137542 7.995705
## 5: 1.759557 -0.059906271 3.984714 3.931298 3.765715 3.813913 3.137542 7.763024
## 6: 1.759557 -0.030455901 3.979339 4.000735 3.819774 3.834219 3.137542 7.866675
## lqre lqfe
## 1: 9.474480 10.85246
## 2: 9.317243 10.69523
## 3: 9.415884 10.79387
## 4: 9.755262 11.13325
## 5: 9.522581 10.90057
## 6: 9.626232 11.00422
str(fdata)
## Classes 'data.table' and 'data.frame': 20000 obs. of 18 variables:
## $ id : int 1 1 1 1 1 1 1 1 1 1 ...
## $ year: int 1 2 3 4 5 6 7 8 9 10 ...
## $ sect: int 1 1 1 1 1 1 1 1 1 1 ...
## $ kg : num 0.00344 0.05239 0.02003 0.02145 -0.00994 ...
## $ lg : num 0.01804 0.01838 0.00141 0.01949 0.01547 ...
## $ mg : num 0.01518 -0.00981 0.00217 0.00514 0.01492 ...
## $ tc : num -0.01161 0.02143 0.01119 0.00726 0.01248 ...
## $ fs : num 3.66 3.66 3.66 3.66 3.66 ...
## $ re : num 1.76 1.76 1.76 1.76 1.76 ...
## $ ee : num 0.18891 -0.05259 -0.00928 0.26721 -0.05991 ...
## $ lk : num 3.68 3.87 3.94 4.02 3.98 ...
## $ ll : num 3.73 3.8 3.8 NA 3.93 ...
## $ lm : num 3.72 3.68 3.69 3.71 3.77 ...
## $ lA : num 3.62 3.7 3.74 3.77 3.81 ...
## $ fe : num 3.14 3.14 3.14 3.14 3.14 ...
## $ lq : num 7.71 7.56 7.66 8 7.76 ...
## $ lqre: num 9.47 9.32 9.42 9.76 9.52 ...
## $ lqfe: num 10.9 10.7 10.8 11.1 10.9 ...
## - attr(*, ".internal.selfref")=<externalptr>
## - attr(*, "index")= int(0)
## ..- attr(*, "__id")= int(0)
summary(fdata)
## id year sect kg
## Min. : 1.0 Min. : 1.00 Min. : 1.0 Min. :-0.061496
## 1st Qu.: 250.8 1st Qu.: 5.75 1st Qu.: 3.0 1st Qu.:-0.003694
## Median : 500.5 Median :10.50 Median : 5.5 Median : 0.009913
## Mean : 500.5 Mean :10.50 Mean : 5.5 Mean : 0.009917
## 3rd Qu.: 750.2 3rd Qu.:15.25 3rd Qu.: 8.0 3rd Qu.: 0.023340
## Max. :1000.0 Max. :20.00 Max. :10.0 Max. : 0.087958
##
## lg mg tc fs
## Min. :-0.027712 Min. :-0.0365834 Min. :-0.029034 Min. :3.665
## 1st Qu.: 0.003926 1st Qu.:-0.0006042 1st Qu.: 0.003148 1st Qu.:3.665
## Median : 0.011012 Median : 0.0060753 Median : 0.009989 Median :3.665
## Mean : 0.010986 Mean : 0.0060809 Mean : 0.009968 Mean :3.665
## 3rd Qu.: 0.018015 3rd Qu.: 0.0127688 3rd Qu.: 0.016758 3rd Qu.:3.665
## Max. : 0.054650 Max. : 0.0438502 Max. : 0.048748 Max. :3.665
##
## re ee lk ll
## Min. :1.76 Min. :-0.358443 Min. :3.214 Min. :3.481
## 1st Qu.:1.76 1st Qu.:-0.065779 1st Qu.:3.802 1st Qu.:3.864
## Median :1.76 Median : 0.000957 Median :4.012 Median :4.070
## Mean :1.76 Mean : 0.001566 Mean :4.050 Mean :4.086
## 3rd Qu.:1.76 3rd Qu.: 0.068347 3rd Qu.:4.250 3rd Qu.:4.282
## Max. :1.76 Max. : 0.381598 Max. :5.557 Max. :5.063
## NA's :955 NA's :988
## lm lA fe lq
## Min. :3.489 Min. :3.562 Min. :2.848 Min. :7.242
## 1st Qu.:3.760 1st Qu.:3.848 1st Qu.:3.122 1st Qu.:7.859
## Median :3.880 Median :4.034 Median :3.222 Median :8.184
## Mean :3.899 Mean :4.050 Mean :3.242 Mean :8.198
## 3rd Qu.:4.016 3rd Qu.:4.227 3rd Qu.:3.345 3rd Qu.:8.511
## Max. :4.671 Max. :4.963 Max. :3.893 Max. :9.566
## NA's :1018
## lqre lqfe
## Min. : 9.001 Min. :10.21
## 1st Qu.: 9.619 1st Qu.:11.09
## Median : 9.944 Median :11.43
## Mean : 9.958 Mean :11.44
## 3rd Qu.:10.271 3rd Qu.:11.76
## Max. :11.325 Max. :13.13
##
table(fdata$year, fdata$sect)
##
## 1 2 3 4 5 6 7 8 9 10
## 1 100 100 100 100 100 100 100 100 100 100
## 2 100 100 100 100 100 100 100 100 100 100
## 3 100 100 100 100 100 100 100 100 100 100
## 4 100 100 100 100 100 100 100 100 100 100
## 5 100 100 100 100 100 100 100 100 100 100
## 6 100 100 100 100 100 100 100 100 100 100
## 7 100 100 100 100 100 100 100 100 100 100
## 8 100 100 100 100 100 100 100 100 100 100
## 9 100 100 100 100 100 100 100 100 100 100
## 10 100 100 100 100 100 100 100 100 100 100
## 11 100 100 100 100 100 100 100 100 100 100
## 12 100 100 100 100 100 100 100 100 100 100
## 13 100 100 100 100 100 100 100 100 100 100
## 14 100 100 100 100 100 100 100 100 100 100
## 15 100 100 100 100 100 100 100 100 100 100
## 16 100 100 100 100 100 100 100 100 100 100
## 17 100 100 100 100 100 100 100 100 100 100
## 18 100 100 100 100 100 100 100 100 100 100
## 19 100 100 100 100 100 100 100 100 100 100
## 20 100 100 100 100 100 100 100 100 100 100
table(fdata$year, fdata$sect)
##
## 1 2 3 4 5 6 7 8 9 10
## 1 100 100 100 100 100 100 100 100 100 100
## 2 100 100 100 100 100 100 100 100 100 100
## 3 100 100 100 100 100 100 100 100 100 100
## 4 100 100 100 100 100 100 100 100 100 100
## 5 100 100 100 100 100 100 100 100 100 100
## 6 100 100 100 100 100 100 100 100 100 100
## 7 100 100 100 100 100 100 100 100 100 100
## 8 100 100 100 100 100 100 100 100 100 100
## 9 100 100 100 100 100 100 100 100 100 100
## 10 100 100 100 100 100 100 100 100 100 100
## 11 100 100 100 100 100 100 100 100 100 100
## 12 100 100 100 100 100 100 100 100 100 100
## 13 100 100 100 100 100 100 100 100 100 100
## 14 100 100 100 100 100 100 100 100 100 100
## 15 100 100 100 100 100 100 100 100 100 100
## 16 100 100 100 100 100 100 100 100 100 100
## 17 100 100 100 100 100 100 100 100 100 100
## 18 100 100 100 100 100 100 100 100 100 100
## 19 100 100 100 100 100 100 100 100 100 100
## 20 100 100 100 100 100 100 100 100 100 100
# Kayıp değerlerin yapısına bakalım
md.pattern(fdata)
## id year sect kg lg mg tc fs re ee lA fe lq lqre lqfe lk ll lm
## 17229 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0
## 865 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1
## 882 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1
## 69 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 2
## 842 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1
## 76 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 2
## 29 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 2
## 8 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 3
## 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 955 988 1018 2961
# Kayıp değerlerin olmadığı bir veri seti hazırlayalım
cdata <- na.omit(fdata)
dim(fdata)
## [1] 20000 18
dim(cdata)
## [1] 17229 18
# Çalışma ortamındaki tüm tabloları görelim
tables()
## NAME NROW NCOL MB COLS KEY
## 1: cdata 17,229 18 2 id,year,sect,kg,lg,mg,...
## 2: fdata 20,000 18 3 id,year,sect,kg,lg,mg,...
## Total: 5MB
datatable işlemleri
İşlemlerin yapısı
- DT[ i, j, by ]
- DT nesnesini kullam
- i de tanımlanan satırlar için,
- j’yi hesapla,
- bu işlemleri by ile tanımlanan her grup için yap
Değişkenlerin silinmesi
# Bir değişken
fdata[, kg := NULL]
# Birden fazla değişken
fdata[, c("lg", "mg", "tc", "fs", "fe", "ee") := NULL]
Satır işlemleri
- Satırlar gözlemleri içerir
# 3-5. satırlar
fdata[3:5,]
## id year sect re lk ll lm lA lq lqre
## 1: 1 3 1 1.759557 3.942558 3.803200 3.692214 3.741553 7.656327 9.415884
## 2: 1 4 1 1.759557 4.021145 NA 3.711040 3.768173 7.995705 9.755262
## 3: 1 5 1 1.759557 3.984714 3.931298 3.765715 3.813913 7.763024 9.522581
## lqfe
## 1: 10.79387
## 2: 11.13325
## 3: 10.90057
fdata[3:5]
## id year sect re lk ll lm lA lq lqre
## 1: 1 3 1 1.759557 3.942558 3.803200 3.692214 3.741553 7.656327 9.415884
## 2: 1 4 1 1.759557 4.021145 NA 3.711040 3.768173 7.995705 9.755262
## 3: 1 5 1 1.759557 3.984714 3.931298 3.765715 3.813913 7.763024 9.522581
## lqfe
## 1: 10.79387
## 2: 11.13325
## 3: 10.90057
# id'si 1 olan tüm satırlar
fdata[id == 1]
## id year sect re lk ll lm lA lq
## 1: 1 1 1 1.759557 3.677162 3.730664 3.720196 3.622019 7.714923
## 2: 1 2 1 1.759557 3.869140 3.798028 3.684258 3.700559 7.557686
## 3: 1 3 1 1.759557 3.942558 3.803200 3.692214 3.741553 7.656327
## 4: 1 4 1 1.759557 4.021145 NA 3.711040 3.768173 7.995705
## 5: 1 5 1 1.759557 3.984714 3.931298 3.765715 3.813913 7.763024
## 6: 1 6 1 1.759557 3.979339 4.000735 3.819774 3.834219 7.866675
## 7: 1 7 1 1.759557 4.101786 4.083978 3.823640 3.853692 7.908060
## 8: 1 8 1 1.759557 4.048504 4.161252 3.822254 3.932113 7.803037
## 9: 1 9 1 1.759557 3.961262 4.166890 3.759391 3.977528 7.972025
## 10: 1 10 1 1.759557 4.068458 4.218723 3.819324 3.993848 7.939336
## 11: 1 11 1 1.759557 4.018583 4.258603 3.810485 4.014686 8.125140
## 12: 1 12 1 1.759557 4.037664 4.284643 3.870614 3.944473 8.009827
## 13: 1 13 1 1.759557 4.181834 4.343615 3.943928 4.006175 8.168516
## 14: 1 14 1 1.759557 4.134559 4.339508 3.939416 4.007519 8.138972
## 15: 1 15 1 1.759557 4.158239 4.370790 3.980651 4.015097 8.286741
## 16: 1 16 1 1.759557 4.192669 4.413958 4.041246 4.001482 8.436208
## 17: 1 17 1 1.759557 4.183384 4.486670 4.127323 4.006634 8.456595
## 18: 1 18 1 1.759557 4.208586 4.538652 4.197928 4.088093 8.684492
## 19: 1 19 1 1.759557 4.286161 4.573148 4.193445 4.125698 8.604524
## 20: 1 20 1 1.759557 4.379317 4.613243 4.160655 4.169695 8.658562
## lqre lqfe
## 1: 9.474480 10.85246
## 2: 9.317243 10.69523
## 3: 9.415884 10.79387
## 4: 9.755262 11.13325
## 5: 9.522581 10.90057
## 6: 9.626232 11.00422
## 7: 9.667617 11.04560
## 8: 9.562594 10.94058
## 9: 9.731582 11.10957
## 10: 9.698893 11.07688
## 11: 9.884697 11.26268
## 12: 9.769385 11.14737
## 13: 9.928073 11.30606
## 14: 9.898530 11.27651
## 15: 10.046298 11.42428
## 16: 10.195765 11.57375
## 17: 10.216152 11.59414
## 18: 10.444049 11.82203
## 19: 10.364081 11.74207
## 20: 10.418119 11.79610
setkey(fdata, id, year)
fdata[list(1)]
## id year sect re lk ll lm lA lq
## 1: 1 1 1 1.759557 3.677162 3.730664 3.720196 3.622019 7.714923
## 2: 1 2 1 1.759557 3.869140 3.798028 3.684258 3.700559 7.557686
## 3: 1 3 1 1.759557 3.942558 3.803200 3.692214 3.741553 7.656327
## 4: 1 4 1 1.759557 4.021145 NA 3.711040 3.768173 7.995705
## 5: 1 5 1 1.759557 3.984714 3.931298 3.765715 3.813913 7.763024
## 6: 1 6 1 1.759557 3.979339 4.000735 3.819774 3.834219 7.866675
## 7: 1 7 1 1.759557 4.101786 4.083978 3.823640 3.853692 7.908060
## 8: 1 8 1 1.759557 4.048504 4.161252 3.822254 3.932113 7.803037
## 9: 1 9 1 1.759557 3.961262 4.166890 3.759391 3.977528 7.972025
## 10: 1 10 1 1.759557 4.068458 4.218723 3.819324 3.993848 7.939336
## 11: 1 11 1 1.759557 4.018583 4.258603 3.810485 4.014686 8.125140
## 12: 1 12 1 1.759557 4.037664 4.284643 3.870614 3.944473 8.009827
## 13: 1 13 1 1.759557 4.181834 4.343615 3.943928 4.006175 8.168516
## 14: 1 14 1 1.759557 4.134559 4.339508 3.939416 4.007519 8.138972
## 15: 1 15 1 1.759557 4.158239 4.370790 3.980651 4.015097 8.286741
## 16: 1 16 1 1.759557 4.192669 4.413958 4.041246 4.001482 8.436208
## 17: 1 17 1 1.759557 4.183384 4.486670 4.127323 4.006634 8.456595
## 18: 1 18 1 1.759557 4.208586 4.538652 4.197928 4.088093 8.684492
## 19: 1 19 1 1.759557 4.286161 4.573148 4.193445 4.125698 8.604524
## 20: 1 20 1 1.759557 4.379317 4.613243 4.160655 4.169695 8.658562
## lqre lqfe
## 1: 9.474480 10.85246
## 2: 9.317243 10.69523
## 3: 9.415884 10.79387
## 4: 9.755262 11.13325
## 5: 9.522581 10.90057
## 6: 9.626232 11.00422
## 7: 9.667617 11.04560
## 8: 9.562594 10.94058
## 9: 9.731582 11.10957
## 10: 9.698893 11.07688
## 11: 9.884697 11.26268
## 12: 9.769385 11.14737
## 13: 9.928073 11.30606
## 14: 9.898530 11.27651
## 15: 10.046298 11.42428
## 16: 10.195765 11.57375
## 17: 10.216152 11.59414
## 18: 10.444049 11.82203
## 19: 10.364081 11.74207
## 20: 10.418119 11.79610
fdata[list(c(1:2)), list(id, year, sect, lq)]
## id year sect lq
## 1: 1 1 1 7.714923
## 2: 1 2 1 7.557686
## 3: 1 3 1 7.656327
## 4: 1 4 1 7.995705
## 5: 1 5 1 7.763024
## 6: 1 6 1 7.866675
## 7: 1 7 1 7.908060
## 8: 1 8 1 7.803037
## 9: 1 9 1 7.972025
## 10: 1 10 1 7.939336
## 11: 1 11 1 8.125140
## 12: 1 12 1 8.009827
## 13: 1 13 1 8.168516
## 14: 1 14 1 8.138972
## 15: 1 15 1 8.286741
## 16: 1 16 1 8.436208
## 17: 1 17 1 8.456595
## 18: 1 18 1 8.684492
## 19: 1 19 1 8.604524
## 20: 1 20 1 8.658562
## 21: 2 1 1 7.395118
## 22: 2 2 1 7.687345
## 23: 2 3 1 7.602643
## 24: 2 4 1 7.659964
## 25: 2 5 1 7.819811
## 26: 2 6 1 7.808144
## 27: 2 7 1 7.728766
## 28: 2 8 1 7.966463
## 29: 2 9 1 7.829543
## 30: 2 10 1 7.899010
## 31: 2 11 1 7.820649
## 32: 2 12 1 8.018926
## 33: 2 13 1 8.088406
## 34: 2 14 1 8.222450
## 35: 2 15 1 8.045352
## 36: 2 16 1 8.407195
## 37: 2 17 1 8.472305
## 38: 2 18 1 8.552817
## 39: 2 19 1 8.611921
## 40: 2 20 1 8.736459
## id year sect lq
fdata[list(c(1:2), c(10,20))]
## id year sect re lk ll lm lA lq lqre
## 1: 1 10 1 1.759557 4.068458 4.218723 3.819324 3.993848 7.939336 9.698893
## 2: 2 20 1 1.759557 4.129917 4.378623 4.030449 4.356764 8.736459 10.496016
## lqfe
## 1: 11.07688
## 2: 12.43439
fdata[(id %in% c(1,2)) & (year %in% c(10, 20))]
## id year sect re lk ll lm lA lq lqre
## 1: 1 10 1 1.759557 4.068458 4.218723 3.819324 3.993848 7.939336 9.698893
## 2: 1 20 1 1.759557 4.379317 4.613243 4.160655 4.169695 8.658562 10.418119
## 3: 2 10 1 1.759557 3.859989 4.137080 3.652591 3.976788 7.899010 9.658567
## 4: 2 20 1 1.759557 4.129917 4.378623 4.030449 4.356764 8.736459 10.496016
## lqfe
## 1: 11.07688
## 2: 11.79610
## 3: 11.59694
## 4: 12.43439
fdata[list(c(1:2))][year %in% c(10, 20)]
## id year sect re lk ll lm lA lq lqre
## 1: 1 10 1 1.759557 4.068458 4.218723 3.819324 3.993848 7.939336 9.698893
## 2: 1 20 1 1.759557 4.379317 4.613243 4.160655 4.169695 8.658562 10.418119
## 3: 2 10 1 1.759557 3.859989 4.137080 3.652591 3.976788 7.899010 9.658567
## 4: 2 20 1 1.759557 4.129917 4.378623 4.030449 4.356764 8.736459 10.496016
## lqfe
## 1: 11.07688
## 2: 11.79610
## 3: 11.59694
## 4: 12.43439
sM <- data.table(id=c(1,1,2,2), year=c(10,20,10,20))
sM
## id year
## 1: 1 10
## 2: 1 20
## 3: 2 10
## 4: 2 20
fdata[sM]
## id year sect re lk ll lm lA lq lqre
## 1: 1 10 1 1.759557 4.068458 4.218723 3.819324 3.993848 7.939336 9.698893
## 2: 1 20 1 1.759557 4.379317 4.613243 4.160655 4.169695 8.658562 10.418119
## 3: 2 10 1 1.759557 3.859989 4.137080 3.652591 3.976788 7.899010 9.658567
## 4: 2 20 1 1.759557 4.129917 4.378623 4.030449 4.356764 8.736459 10.496016
## lqfe
## 1: 11.07688
## 2: 11.79610
## 3: 11.59694
## 4: 12.43439
Sütun işlemler,
- Sütunlarda değişkenler yer alır
# id sütununun 3-5. satırları
fdata[3:5, id]
## [1] 1 1 1
# id, year ve lq sütunlarının 3-5. satırları
fdata[3:5, list(id, year, lq)]
## id year lq
## 1: 1 3 7.656327
## 2: 1 4 7.995705
## 3: 1 5 7.763024
# NOT: .() list() anlamına gelmektedir
# lk'nin kayıp değerde olduğu gözlemlerin ilk 10 satırı (id, year ve lq değişkenleri)
fdata[is.na(lk), .(id, year, lq, lk)][1:10]
## id year lq lk
## 1: 4 11 8.241903 NA
## 2: 5 11 8.291793 NA
## 3: 8 1 7.655521 NA
## 4: 8 2 7.644491 NA
## 5: 8 3 7.686605 NA
## 6: 8 4 7.670166 NA
## 7: 8 5 7.797126 NA
## 8: 8 6 7.693435 NA
## 9: 8 7 7.662222 NA
## 10: 8 8 7.818517 NA
# lq'nin ortalaması
fdata[, mean(lq)]
## [1] 8.198103
# lq'nin ortalaması ve standart hatası
fdata[, .(mean(lq), sd(lq))]
## V1 V2
## 1: 8.198103 0.4161142
# lk'nin ortalaması ve standart hatası
fdata[, .(mean(lk), sd(lk))]
## V1 V2
## 1: NA NA
# Yeni değişken yaratmak
fdata[, list(mean.lq = mean(lq), sd.lq = sd(lq))]
## mean.lq sd.lq
## 1: 8.198103 0.4161142
# datatable işlemleri olarak fonksiyonlar da kullanılabilir
# do-loop'a gerek yok!
fdata[, {plot(lq)
plot(ll)}]
## NULL
Grup bazında işlemler
# Grup işlemleri
fdata[, list(mean.lq = mean(lq), sd.lq = sd(lq)), by=sect]
## sect mean.lq sd.lq
## 1: 1 8.180033 0.4082513
## 2: 2 8.200705 0.4193998
## 3: 3 8.194084 0.4190149
## 4: 4 8.214795 0.4258512
## 5: 5 8.185673 0.4028809
## 6: 6 8.190917 0.4099462
## 7: 7 8.189549 0.4106374
## 8: 8 8.193422 0.4184685
## 9: 9 8.240427 0.4299026
## 10: 10 8.191424 0.4136558
# Birden fazla değişken ile gruplama
sectq <- fdata[, list(mean.lq = mean(lq), sd.lq = sd(lq)), by=list(sect, year)]
sectq
## sect year mean.lq sd.lq
## 1: 1 1 7.568505 0.1127734
## 2: 1 2 7.656346 0.1119503
## 3: 1 3 7.724386 0.1114008
## 4: 1 4 7.750268 0.1488633
## 5: 1 5 7.846275 0.1708398
## ---
## 196: 10 16 8.546901 0.2147203
## 197: 10 17 8.620089 0.2005785
## 198: 10 18 8.670703 0.2500211
## 199: 10 19 8.722522 0.2196667
## 200: 10 20 8.781971 0.2416809
tables()
## NAME NROW NCOL MB COLS KEY
## 1: cdata 17,229 18 2 id,year,sect,kg,lg,mg,...
## 2: fdata 20,000 11 1 id,year,sect,re,lk,ll,... id,year
## 3: sectq 200 4 0 sect,year,mean.lq,sd.lq
## 4: sM 4 2 0 id,year
## Total: 3MB
# by içinde operatör/fınksiyon kullanılması
fdata[, list(mean.lq = mean(lq), sd.lq = sd(lq)), by=list(sect, year>5)]
## sect year mean.lq sd.lq
## 1: 1 FALSE 7.709156 0.1622812
## 2: 1 TRUE 8.336992 0.3389476
## 3: 2 FALSE 7.707978 0.1620525
## 4: 2 TRUE 8.364947 0.3432946
## 5: 3 FALSE 7.704858 0.1619919
## 6: 3 TRUE 8.357159 0.3449004
## 7: 4 FALSE 7.703277 0.1636964
## 8: 4 TRUE 8.385301 0.3413980
## 9: 5 FALSE 7.715725 0.1554984
## 10: 5 TRUE 8.342322 0.3319376
## 11: 6 FALSE 7.714602 0.1548702
## 12: 6 TRUE 8.349688 0.3394505
## 13: 7 FALSE 7.711210 0.1558822
## 14: 7 TRUE 8.348995 0.3391443
## 15: 8 FALSE 7.707212 0.1547212
## 16: 8 TRUE 8.355492 0.3470176
## 17: 9 FALSE 7.723819 0.1553899
## 18: 9 TRUE 8.412629 0.3460214
## 19: 10 FALSE 7.700174 0.1623720
## 20: 10 TRUE 8.355174 0.3347864
fdata[, list(mean.lq = mean(lq), sd.lq = sd(lq), nobs = length(lq)), by=is.na(ll)]
## is.na mean.lq sd.lq nobs
## 1: FALSE 8.197396 0.4151473 19012
## 2: TRUE 8.211708 0.4342968 988
# Seçilen satırlar için gruplama
fdata[1:5, list(mean.lq = mean(lq), sd.lq = sd(lq)), by=year]
## year mean.lq sd.lq
## 1: 1 7.714923 NA
## 2: 2 7.557686 NA
## 3: 3 7.656327 NA
## 4: 4 7.995705 NA
## 5: 5 7.763024 NA
# .N ile kayıp gözlem sayısının hesaplanması
fdata[is.na(lk), .N, by = sect]
## sect N
## 1: 1 146
## 2: 2 78
## 3: 3 45
## 4: 4 82
## 5: 5 62
## 6: 6 111
## 7: 7 85
## 8: 8 141
## 9: 9 72
## 10: 10 133
Yeni değişken yaratılması
head(fdata)
## id year sect re lk ll lm lA lq lqre
## 1: 1 1 1 1.759557 3.677162 3.730664 3.720196 3.622019 7.714923 9.474480
## 2: 1 2 1 1.759557 3.869140 3.798028 3.684258 3.700559 7.557686 9.317243
## 3: 1 3 1 1.759557 3.942558 3.803200 3.692214 3.741553 7.656327 9.415884
## 4: 1 4 1 1.759557 4.021145 NA 3.711040 3.768173 7.995705 9.755262
## 5: 1 5 1 1.759557 3.984714 3.931298 3.765715 3.813913 7.763024 9.522581
## 6: 1 6 1 1.759557 3.979339 4.000735 3.819774 3.834219 7.866675 9.626232
## lqfe
## 1: 10.85246
## 2: 10.69523
## 3: 10.79387
## 4: 11.13325
## 5: 10.90057
## 6: 11.00422
fdata[, Q1 := exp(lq)]
head(fdata)
## id year sect re lk ll lm lA lq lqre
## 1: 1 1 1 1.759557 3.677162 3.730664 3.720196 3.622019 7.714923 9.474480
## 2: 1 2 1 1.759557 3.869140 3.798028 3.684258 3.700559 7.557686 9.317243
## 3: 1 3 1 1.759557 3.942558 3.803200 3.692214 3.741553 7.656327 9.415884
## 4: 1 4 1 1.759557 4.021145 NA 3.711040 3.768173 7.995705 9.755262
## 5: 1 5 1 1.759557 3.984714 3.931298 3.765715 3.813913 7.763024 9.522581
## 6: 1 6 1 1.759557 3.979339 4.000735 3.819774 3.834219 7.866675 9.626232
## lqfe Q1
## 1: 10.85246 2241.550
## 2: 10.69523 1915.408
## 3: 10.79387 2113.979
## 4: 11.13325 2968.181
## 5: 10.90057 2352.006
## 6: 11.00422 2608.877
# Yeni değişken yaratıldığında DT <- DT[...] işlemi gerekli değil!
# := yeni değişkenleri DT içinde oluşturur
fdata[, c("K", "L", "M") := list(exp(lk), exp(ll), exp(lm))]
head(fdata)
## id year sect re lk ll lm lA lq lqre
## 1: 1 1 1 1.759557 3.677162 3.730664 3.720196 3.622019 7.714923 9.474480
## 2: 1 2 1 1.759557 3.869140 3.798028 3.684258 3.700559 7.557686 9.317243
## 3: 1 3 1 1.759557 3.942558 3.803200 3.692214 3.741553 7.656327 9.415884
## 4: 1 4 1 1.759557 4.021145 NA 3.711040 3.768173 7.995705 9.755262
## 5: 1 5 1 1.759557 3.984714 3.931298 3.765715 3.813913 7.763024 9.522581
## 6: 1 6 1 1.759557 3.979339 4.000735 3.819774 3.834219 7.866675 9.626232
## lqfe Q1 K L M
## 1: 10.85246 2241.550 39.53405 41.70678 41.27247
## 2: 10.69523 1915.408 47.90119 44.61312 39.81558
## 3: 10.79387 2113.979 51.55027 44.84447 40.13360
## 4: 11.13325 2968.181 55.76489 NA 40.89631
## 5: 10.90057 2352.006 53.76991 50.97312 43.19459
## 6: 11.00422 2608.877 53.48165 54.63832 45.59392
Endeksleme ve anahtarlar
setkey(fdata, id, year)
fdata[.(1), .(id, year, Q1)]
## id year Q1
## 1: 1 1 2241.550
## 2: 1 2 1915.408
## 3: 1 3 2113.979
## 4: 1 4 2968.181
## 5: 1 5 2352.006
## 6: 1 6 2608.877
## 7: 1 7 2719.109
## 8: 1 8 2448.024
## 9: 1 9 2898.721
## 10: 1 10 2805.497
## 11: 1 11 3378.340
## 12: 1 12 3010.398
## 13: 1 13 3528.104
## 14: 1 14 3425.396
## 15: 1 15 3970.871
## 16: 1 16 4611.036
## 17: 1 17 4706.006
## 18: 1 18 5910.536
## 19: 1 19 5456.289
## 20: 1 20 5759.246
fdata[.(c(1:10)), mult = "first", .(id, year, Q1)]
## id year Q1
## 1: 1 1 2241.550
## 2: 2 1 1628.017
## 3: 3 1 2165.505
## 4: 4 1 1715.678
## 5: 5 1 1691.209
## 6: 6 1 1707.853
## 7: 7 1 1840.785
## 8: 8 1 2112.276
## 9: 9 1 2536.401
## 10: 10 1 2100.196
fdata[.(c(1:10)), mult = "last", .(id, year, Q1)]
## id year Q1
## 1: 1 20 5759.246
## 2: 2 20 6225.813
## 3: 3 20 7385.818
## 4: 4 20 6731.757
## 5: 5 20 8693.780
## 6: 6 20 6411.944
## 7: 7 20 7293.666
## 8: 8 20 8070.181
## 9: 9 20 11502.584
## 10: 10 20 6146.992
fdata[.(1,20)]
## id year sect re lk ll lm lA lq lqre
## 1: 1 20 1 1.759557 4.379317 4.613243 4.160655 4.169695 8.658562 10.41812
## lqfe Q1 K L M
## 1: 11.7961 5759.246 79.7835 100.8106 64.11348
fdata[.(c(1:10)), sum(Q1), by = .EACHI]
## id V1
## 1: 1 68827.58
## 2: 2 65040.03
## 3: 3 83227.48
## 4: 4 75885.58
## 5: 5 92356.68
## 6: 6 75939.10
## 7: 7 84751.83
## 8: 8 73023.61
## 9: 9 96708.35
## 10: 10 73910.11
Değişken ismi değiştirilmesi
setnames(fdata, "Q1", "Q")
# Değişken ismini değiştir
fdata[, list(sum(Q), round(sum(L, na.rm = TRUE), 0)), by = year]
## year V1 V2
## 1: 1 1962948 38522
## 2: 2 2105230 40088
## 3: 3 2246275 41785
## 4: 4 2400973 43720
## 5: 5 2578635 45726
## 6: 6 2740260 46761
## 7: 7 2946779 49223
## 8: 8 3136444 51582
## 9: 9 3374734 53495
## 10: 10 3592687 55608
## 11: 11 3852486 58316
## 12: 12 4105420 61087
## 13: 13 4345016 62950
## 14: 14 4626340 66019
## 15: 15 4971910 68807
## 16: 16 5284673 71770
## 17: 17 5685954 74776
## 18: 18 6039388 77633
## 19: 19 6465543 81118
## 20: 20 6906750 83643
# Ve sırala
fdata[order(-year)]
## id year sect re lk ll lm lA lq
## 1: 1 20 1 1.759557 4.379317 4.613243 4.160655 4.169695 8.658562
## 2: 2 20 1 1.759557 4.129917 4.378623 4.030449 4.356764 8.736459
## 3: 3 20 1 1.759557 4.043933 4.242055 4.285151 4.344415 8.907317
## 4: 4 20 1 1.759557 4.724215 4.473943 4.052891 4.437002 8.814591
## 5: 5 20 1 1.759557 3.960350 4.607179 4.028273 4.647148 9.070363
## ---
## 19996: 996 1 10 1.759557 3.606828 3.690967 3.725613 3.732710 7.684911
## 19997: 997 1 10 1.759557 3.787980 3.643523 3.727933 3.614846 7.635785
## 19998: 998 1 10 1.759557 3.700249 3.624219 3.659327 3.642017 7.488636
## 19999: 999 1 10 1.759557 3.705054 3.747131 3.680591 3.679624 7.525502
## 20000: 1000 1 10 1.759557 3.762371 3.680319 NA 3.710717 7.470778
## lqre lqfe Q K L M
## 1: 10.418119 11.79610 5759.246 79.78350 100.81059 64.11348
## 2: 10.496016 12.43439 6225.813 62.17276 79.72818 56.28619
## 3: 10.666874 12.09744 7385.818 57.05029 69.55063 72.61354
## 4: 10.574149 12.18435 6731.757 112.64198 87.70183 57.56362
## 5: 10.829920 12.22065 8693.780 52.47571 100.20108 56.16382
## ---
## 19996: 9.444468 11.07915 2175.275 36.84900 40.08358 41.49665
## 19997: 9.395342 10.83997 2070.995 44.16711 38.22628 41.59306
## 19998: 9.248194 10.67783 1787.613 40.45739 37.49543 38.83520
## 19999: 9.285059 10.76246 1854.744 40.65225 42.39926 39.66984
## 20000: 9.230335 10.77457 1755.972 43.05040 39.65905 NA
Zincirleme işlemler
# İki işlemi aynı anda yap
fdata[, .(sum(Q), round(sum(L, na.rm = TRUE), 0)), by = year][order(-year)]
## year V1 V2
## 1: 20 6906750 83643
## 2: 19 6465543 81118
## 3: 18 6039388 77633
## 4: 17 5685954 74776
## 5: 16 5284673 71770
## 6: 15 4971910 68807
## 7: 14 4626340 66019
## 8: 13 4345016 62950
## 9: 12 4105420 61087
## 10: 11 3852486 58316
## 11: 10 3592687 55608
## 12: 9 3374734 53495
## 13: 8 3136444 51582
## 14: 7 2946779 49223
## 15: 6 2740260 46761
## 16: 5 2578635 45726
## 17: 4 2400973 43720
## 18: 3 2246275 41785
## 19: 2 2105230 40088
## 20: 1 1962948 38522
Verilerin şekillendirilmesi
- Verilerin eritilmesi (melting): Uzun format
- Verilerin kalıba dökülmesi (casting): Geniş format
# Sektörel fiyatları oluşturalım
sPrice <- data.table(
year = c(1:nt),
p1 = 1 + cumsum(rnorm(nt, 0.02, 0.04)),
p2 = 1 + cumsum(rnorm(nt, 0.03, 0.05)),
p3 = 1 + cumsum(rnorm(nt, 0.04, 0.02)),
p4 = 1 + cumsum(rnorm(nt, 0.05, 0.03)),
p5 = 1 + cumsum(rnorm(nt, 0.02, 0.04)),
p6 = 1 + cumsum(rnorm(nt, 0.03, 0.06)),
p7 = 1 + cumsum(rnorm(nt, 0.04, 0.02)),
p8 = 1 + cumsum(rnorm(nt, 0.05, 0.07)),
p9 = 1 + cumsum(rnorm(nt, 0.05, 0.04)),
p10 = 1 + cumsum(rnorm(nt, 0.05, 0.06)))
head(sPrice)
## year p1 p2 p3 p4 p5 p6 p7 p8
## 1: 1 1.084834 1.086403 1.044239 1.039482 1.120569 1.083191 1.036712 1.093676
## 2: 2 1.101266 1.166263 1.087915 1.066618 1.242368 1.160425 1.069804 1.201331
## 3: 3 1.164072 1.226919 1.125465 1.120647 1.222590 1.223205 1.112062 1.288835
## 4: 4 1.164269 1.318204 1.136977 1.167959 1.200625 1.335707 1.147675 1.337262
## 5: 5 1.200572 1.355829 1.178433 1.204771 1.217625 1.290941 1.206488 1.412035
## 6: 6 1.244234 1.358888 1.192575 1.299423 1.309964 1.434764 1.258097 1.402196
## p9 p10
## 1: 1.093882 1.0195968
## 2: 1.161360 1.0470296
## 3: 1.185696 0.9921116
## 4: 1.220048 1.1548797
## 5: 1.296979 1.2125423
## 6: 1.350105 1.2534466
sPriceM <- melt(sPrice, id.vars=1, variable.name = "sectName", value.name = "price")
head(sPriceM)
## year sectName price
## 1: 1 p1 1.084834
## 2: 2 p1 1.101266
## 3: 3 p1 1.164072
## 4: 4 p1 1.164269
## 5: 5 p1 1.200572
## 6: 6 p1 1.244234
sPriceM$sect <- as.numeric(unlist(gsub("[^0-9]", "", unlist(sPriceM$sectName)), ""))
sPriceM[, sectName := NULL]
head(sPriceM)
## year price sect
## 1: 1 1.084834 1
## 2: 2 1.101266 1
## 3: 3 1.164072 1
## 4: 4 1.164269 1
## 5: 5 1.200572 1
## 6: 6 1.244234 1
ggplot(sPriceM, aes(x=year, y=price, group=sect, color=as.factor(sect))) + geom_line() +
ggtitle("Price indices")
setkey(sPriceM, sect, year)
setkey(fdata, sect, year)
fdataM <- fdata[sPriceM]
head(fdataM)
## id year sect re lk ll lm lA lq lqre
## 1: 1 1 1 1.759557 3.677162 3.730664 3.720196 3.622019 7.714923 9.474480
## 2: 2 1 1 1.759557 3.771031 3.719085 3.693771 3.704430 7.395118 9.154675
## 3: 3 1 1 1.759557 3.779838 3.614786 3.706404 3.752991 7.680409 9.439966
## 4: 4 1 1 1.759557 3.704125 3.617248 3.570059 3.709631 7.447564 9.207121
## 5: 5 1 1 1.759557 3.619427 3.663226 3.716082 3.730261 7.433199 9.192756
## 6: 6 1 1 1.759557 3.861231 NA 3.612980 3.718183 7.442992 9.202550
## lqfe Q K L M price
## 1: 10.85246 2241.550 39.53405 41.70678 41.27247 1.084834
## 2: 11.09305 1628.017 43.42479 41.22665 40.19616 1.084834
## 3: 10.87053 2165.505 43.80896 37.14340 40.70714 1.084834
## 4: 10.81732 1715.678 40.61450 37.23495 35.51868 1.084834
## 5: 10.58349 1691.209 37.31618 38.98692 41.10305 1.084834
## 6: 10.63054 1707.853 47.52379 NA 37.07637 1.084834
Veri setlerinib birleştirilmesi
- Sektör düzeyinde fiyatları oluşturalım
sdata <- fdata[, .(sQ = sum(Q, na.rm=T), sL = sum(L, na.rm=T), sM = sum(M, na.rm=T)),
by = .(sect, year)]
summary(sdata)
## sect year sQ sL sM
## Min. : 1.0 Min. : 1.00 Min. :194352 Min. :3730 Min. :3616
## 1st Qu.: 3.0 1st Qu.: 5.75 1st Qu.:267073 1st Qu.:4620 1st Qu.:4216
## Median : 5.5 Median :10.50 Median :370984 Median :5692 Median :4742
## Mean : 5.5 Mean :10.50 Mean :396842 Mean :5863 Mean :4756
## 3rd Qu.: 8.0 3rd Qu.:15.25 3rd Qu.:515701 3rd Qu.:7037 3rd Qu.:5272
## Max. :10.0 Max. :20.00 Max. :748102 Max. :8758 Max. :6069
dim(sdata)
## [1] 200 5
# Sektörel fiyatları firma verileri ile birleştir
setkey(fdata, sect, year)
setkey(sdata, sect, year)
hdata <- fdata[sdata]
dim(hdata)
## [1] 20000 18
dim(fdata)
## [1] 20000 15
identical(hdata$ll, fdata$ll)
## [1] TRUE
# Daha kısa yol
fdata[, c("sQ", "sL", "sM") := list(sum(Q, na.rm=T), sum(L, na.rm=T), sum(M, na.rm=T)),
by = .(sect, year)]
identical(hdata, fdata)
## [1] FALSE
# Sektörel geometrik ortalamalardan farkı bul
fdata[, lqdev := lq - mean(lq, na.rm=T), by = .(sect)]
fdata[, lkdev := lk - mean(lk, na.rm=T), by = .(sect)]
fdata[, lldev := ll - mean(ll, na.rm=T), by = .(sect)]
fdata[, lmdev := lm - mean(lm, na.rm=T), by = .(sect)]
Regresyon analizi
Model ve tahmin
model1 <- lm(lq ~ lk + ll + lm + year, data = fdata)
model2 <- lm(lqdev ~ lkdev + lldev + lmdev + year, data = fdata)
model3 <- lm(lqre ~ lk + ll + lm + year, data = fdata)
model4 <- lm(lqfe ~ lk + ll + lm + year, data = fdata)
model5 <- plm(lq ~ lk + ll + lm + year, data = fdata, model="random", effect="individual")
model6 <- plm(lq ~ lk + ll + lm + year, data = fdata, model="within", effect="individual")
stargazer(model1, model2, model3, model4, model5, model6, type="html",
model.numbers=FALSE,
dep.var.labels.include = FALSE, dep.var.caption = "Production function estimates",
column.labels = c("OLS", "OLS+dev", "OLS RE", "OLS FE", "Random effects", "Fixed effects"))
|
|
|
|
Production function estimates
|
|
|
|
|
|
OLS
|
OLS
|
OLS
|
OLS
|
panel
|
|
|
|
|
|
|
linear
|
|
|
OLS
|
OLS+dev
|
OLS RE
|
OLS FE
|
Random effects
|
Fixed effects
|
|
|
|
lk
|
0.112***
|
|
0.112***
|
0.246***
|
0.110***
|
0.110***
|
|
|
(0.005)
|
|
(0.005)
|
(0.008)
|
(0.007)
|
(0.007)
|
|
|
|
|
|
|
|
|
|
ll
|
0.198***
|
|
0.198***
|
0.366***
|
0.214***
|
0.218***
|
|
|
(0.010)
|
|
(0.010)
|
(0.014)
|
(0.013)
|
(0.014)
|
|
|
|
|
|
|
|
|
|
lm
|
0.745***
|
|
0.745***
|
0.684***
|
0.762***
|
0.766***
|
|
|
(0.011)
|
|
(0.011)
|
(0.015)
|
(0.013)
|
(0.014)
|
|
|
|
|
|
|
|
|
|
lkdev
|
|
0.112***
|
|
|
|
|
|
|
|
(0.005)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
lldev
|
|
0.196***
|
|
|
|
|
|
|
|
(0.010)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
lmdev
|
|
0.744***
|
|
|
|
|
|
|
|
(0.011)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
year
|
0.036***
|
0.036***
|
0.036***
|
0.026***
|
|
|
|
|
(0.0005)
|
(0.0005)
|
(0.0005)
|
(0.001)
|
|
|
|
|
|
|
|
|
|
|
|
year2
|
|
|
|
|
0.037***
|
0.037***
|
|
|
|
|
|
|
(0.006)
|
(0.006)
|
|
|
|
|
|
|
|
|
|
year3
|
|
|
|
|
0.070***
|
0.069***
|
|
|
|
|
|
|
(0.006)
|
(0.006)
|
|
|
|
|
|
|
|
|
|
year4
|
|
|
|
|
0.105***
|
0.104***
|
|
|
|
|
|
|
(0.006)
|
(0.006)
|
|
|
|
|
|
|
|
|
|
year5
|
|
|
|
|
0.149***
|
0.148***
|
|
|
|
|
|
|
(0.006)
|
(0.006)
|
|
|
|
|
|
|
|
|
|
year6
|
|
|
|
|
0.175***
|
0.174***
|
|
|
|
|
|
|
(0.007)
|
(0.007)
|
|
|
|
|
|
|
|
|
|
year7
|
|
|
|
|
0.220***
|
0.219***
|
|
|
|
|
|
|
(0.007)
|
(0.007)
|
|
|
|
|
|
|
|
|
|
year8
|
|
|
|
|
0.251***
|
0.249***
|
|
|
|
|
|
|
(0.007)
|
(0.007)
|
|
|
|
|
|
|
|
|
|
year9
|
|
|
|
|
0.289***
|
0.288***
|
|
|
|
|
|
|
(0.008)
|
(0.008)
|
|
|
|
|
|
|
|
|
|
year10
|
|
|
|
|
0.320***
|
0.318***
|
|
|
|
|
|
|
(0.008)
|
(0.008)
|
|
|
|
|
|
|
|
|
|
year11
|
|
|
|
|
0.359***
|
0.357***
|
|
|
|
|
|
|
(0.008)
|
(0.009)
|
|
|
|
|
|
|
|
|
|
year12
|
|
|
|
|
0.394***
|
0.391***
|
|
|
|
|
|
|
(0.009)
|
(0.009)
|
|
|
|
|
|
|
|
|
|
year13
|
|
|
|
|
0.419***
|
0.416***
|
|
|
|
|
|
|
(0.009)
|
(0.010)
|
|
|
|
|
|
|
|
|
|
year14
|
|
|
|
|
0.454***
|
0.451***
|
|
|
|
|
|
|
(0.010)
|
(0.010)
|
|
|
|
|
|
|
|
|
|
year15
|
|
|
|
|
0.499***
|
0.495***
|
|
|
|
|
|
|
(0.010)
|
(0.011)
|
|
|
|
|
|
|
|
|
|
year16
|
|
|
|
|
0.526***
|
0.523***
|
|
|
|
|
|
|
(0.010)
|
(0.011)
|
|
|
|
|
|
|
|
|
|
year17
|
|
|
|
|
0.567***
|
0.563***
|
|
|
|
|
|
|
(0.011)
|
(0.012)
|
|
|
|
|
|
|
|
|
|
year18
|
|
|
|
|
0.596***
|
0.592***
|
|
|
|
|
|
|
(0.011)
|
(0.012)
|
|
|
|
|
|
|
|
|
|
year19
|
|
|
|
|
0.634***
|
0.630***
|
|
|
|
|
|
|
(0.012)
|
(0.013)
|
|
|
|
|
|
|
|
|
|
year20
|
|
|
|
|
0.668***
|
0.663***
|
|
|
|
|
|
|
(0.012)
|
(0.013)
|
|
|
|
|
|
|
|
|
|
Constant
|
3.651***
|
-0.380***
|
5.410***
|
6.008***
|
3.568***
|
|
|
|
(0.051)
|
(0.005)
|
(0.051)
|
(0.075)
|
(0.066)
|
|
|
|
|
|
|
|
|
|
|
|
|
Observations
|
17,229
|
17,229
|
17,229
|
17,229
|
17,229
|
17,229
|
|
R2
|
0.859
|
0.860
|
0.859
|
0.746
|
0.908
|
0.910
|
|
Adjusted R2
|
0.859
|
0.860
|
0.859
|
0.746
|
0.908
|
0.905
|
|
Residual Std. Error (df = 17224)
|
0.156
|
0.156
|
0.156
|
0.228
|
|
|
|
F Statistic
|
26,201.980*** (df = 4; 17224)
|
26,367.730*** (df = 4; 17224)
|
26,201.980*** (df = 4; 17224)
|
12,634.110*** (df = 4; 17224)
|
164,933.400***
|
7,470.656*** (df = 22; 16278)
|
|
|
|
Note:
|
p<0.1; p<0.05; p<0.01
|
Regresyon sonuçları
attributes(model1)
## $names
## [1] "coefficients" "residuals" "effects" "rank"
## [5] "fitted.values" "assign" "qr" "df.residual"
## [9] "na.action" "xlevels" "call" "terms"
## [13] "model"
##
## $class
## [1] "lm"
attributes(model5)
## $names
## [1] "coefficients" "vcov" "residuals" "df.residual" "formula"
## [6] "model" "ercomp" "assign" "contrasts" "args"
## [11] "aliased" "call"
##
## $class
## [1] "plm" "panelmodel"
class(model1$residuals)
## [1] "numeric"
hist(model1$residuals, plot = TRUE)
Monte Carlo analizi
- Örneklem büyüklüğü ve tahmin hatası arasındaki ilişki
ntest <- 1000
coeffA <- matrix(NA, nrow=ntest, ncol=5)
coeffB <- matrix(NA, nrow=ntest, ncol=5)
for (j in c(1:5)) {
for (i in c(1:ntest)) {
ns <- floor(10*exp(j))
x1 <- rnorm(ns, 0, 1)
x2 <- rnorm(ns, 0, 2)
y <- 1 + 0.5*x1 + 2*x2 + rnorm(ns, 0, 1)
mod <- lm(y ~ x1 + x2)
coeffA[i,j] <- mod$coefficients[2]
coeffB[i,j] <- mod$coefficients[3]
}
}
coeffA <- as.data.frame(coeffA)
coeffB <- as.data.frame(coeffB)
names(coeffB) <- names(coeffA) <- as.character(floor(10*exp(c(1:5))))
head(coeffA)
## 27 73 200 545 1484
## 1 0.2922516 0.7159659 0.3938043 0.5375261 0.4639097
## 2 0.4407526 0.4284639 0.4740828 0.5068020 0.4948005
## 3 0.6378205 0.6418433 0.5092009 0.5826468 0.4872815
## 4 0.3617704 0.3067727 0.4320805 0.4950480 0.4766151
## 5 0.3642107 0.5168827 0.4406001 0.5524684 0.4907939
## 6 0.1774476 0.4594442 0.5708598 0.5133530 0.4686676
coeffAA <- melt(coeffA, id.vars=0, variable.name = "N_obs", value.name = "Estimate")
## Warning in melt(coeffA, id.vars = 0, variable.name = "N_obs", value.name =
## "Estimate"): The melt generic in data.table has been passed a data.frame and
## will attempt to redirect to the relevant reshape2 method; please note that
## reshape2 is deprecated, and this redirection is now deprecated as well. To
## continue using melt methods from reshape2 while both libraries are attached,
## e.g. melt.list, you can prepend the namespace like reshape2::melt(coeffA). In
## the next version, this warning will become an error.
head(coeffAA)
## N_obs Estimate
## 1 27 0.2922516
## 2 27 0.4407526
## 3 27 0.6378205
## 4 27 0.3617704
## 5 27 0.3642107
## 6 27 0.1774476
coeffBB <- melt(coeffB, id.vars=0, variable.name = "N_obs", value.name = "Estimate")
## Warning in melt(coeffB, id.vars = 0, variable.name = "N_obs", value.name =
## "Estimate"): The melt generic in data.table has been passed a data.frame and
## will attempt to redirect to the relevant reshape2 method; please note that
## reshape2 is deprecated, and this redirection is now deprecated as well. To
## continue using melt methods from reshape2 while both libraries are attached,
## e.g. melt.list, you can prepend the namespace like reshape2::melt(coeffB). In
## the next version, this warning will become an error.
ggplot(coeffAA, aes(x=Estimate, group=N_obs, color=N_obs)) + geom_density() +
ggtitle("Coefficient A")
ggplot(coeffBB, aes(x=Estimate, group=N_obs, color=N_obs)) + geom_density() +
ggtitle("Coefficient B")
