API tools faq
paste
Advertisement
Guest User

Untitled

a guest
Jul 27th, 2017
124
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 11.18 KB | None | 0 0
raw download clone embed print report
  1. # R code written by: Isidora Gatarić
  2. # More about experiment: Gatarić, I., & Filipović Đurđević, D. (2016). Processing effects of semantic ambiguity of Serbian derivational # suffixes. Paper presented at the 17th International Morphology Meeting, Vienna, Austria. Book of abstracts, pp. 45-46.
  3.  
  4. require(rms)
  5. require(lme4)
  6. require(languageR)
  7. require(multcomp)
  8. require(lsmeans)
  9. require(multcompView)
  10. require(pbkrtest)
  11. require(lmerTest)
  12. require(randomForest)
  13. require(party)
  14. require(MASS)
  15. require(ggplot2)
  16. library(car)
  17. require(itsadug)
  18.  
  19. dat=read.table("vldfinalno.txt",, sep="\t",header=TRUE)
  20.  
  21. dim(dat)
  22. #[1] 2023 18
  23.  
  24. colnames(dat)
  25.  
  26. # [1] "Ispitanik" "TrialOrder" "Grupa"
  27. # [4] "CorrectResponse" "TrialNumber" "Sufiks"
  28. # [7] "DuzinaSufiksa" "Response" "Tacnost"
  29. #[10] "RT" "NV" "Stimulus"
  30. #[13] "FrekLemKostic" "PovFrekWac" "FrekLemWac"
  31. #[16] "DuzinaReci" "FrekvencijaSufiksa" "ProduktivnostSufiksa"
  32.  
  33. _____________________________________________
  34. __________________________________________________________
  35.  
  36. # Vizulena inspekcija podataka, grafik VIP1
  37.  
  38. par(mfrow=c(2,2))
  39. plot(sort(dat$RT))
  40. plot(density(dat$RT))
  41. qqnorm(dat$RT)
  42. par(mfrow=c(1,1))
  43.  
  44. powerTransform(dat$RT)
  45. #Estimated transformation parameters
  46. # dat$RT
  47. #-1.314563
  48.  
  49. # Log frekvencija leme, frekvencija sufiksa, duzina sufiksa i duzina leme
  50.  
  51. dat$dlem=log(dat$DuzinaReci)
  52. dat$flemk=log(dat$FrekLemKostic)
  53. dat$flemw=log(dat$FrekLemWac)
  54. dat$fpov=log(dat$PovFrekWac)
  55. dat$fsuf=log(dat$FrekvencijaSufiksa)
  56. dat$dsuf=log(dat$DuzinaSufiksa)
  57. dat$psuf=log(dat$ProduktivnostSufiksa)
  58.  
  59. # Normalizujem kontinuirane prediktore
  60.  
  61. dat$trial.z = scale(dat$TrialOrder)
  62. dat$len.z = scale(dat$dlem)
  63. dat$flemk.z=scale(dat$flemk)
  64. dat$flemw.z=scale(dat$flemw)
  65. dat$fpov.z=scale(dat$fpov)
  66. dat$fsuf.z = scale(dat$fsuf)
  67. dat$dsuf.z = scale(dat$dsuf)
  68. dat$psuf.z = scale(dat$psuf)
  69.  
  70. # Pobrinem se da je faktor tretiran kao faktor
  71.  
  72. as.factor(as.character(dat$NV))
  73. levels(dat$NV)
  74. table(dat$NV)
  75.  
  76. as.factor(as.character(dat$Ispitanik))
  77. levels(dat$Ispitanik)
  78. table(dat$Ispitanik)
  79.  
  80. as.factor(as.character(dat$Stimulus))
  81. levels(dat$Stimulus)
  82. table(dat$Stimulus)
  83. __________________________________________________________
  84. ________________________________________________________________
  85.  
  86. #Kontinuirane prediktore -- da vidim kako vizuelno to sve izgleda
  87.  
  88. -------------------------
  89. #Trial Order
  90. -------------------------
  91.  
  92. par(mfrow=c(2,2))
  93. plot(sort(dat$TrialOrder))
  94. plot(density(dat$TrialOrder))
  95. qqnorm(dat$TrialOrder)
  96. par(mfrow=c(1,1))
  97.  
  98. -------------------------
  99. #Duzina Reci
  100. -------------------------
  101.  
  102. par(mfrow=c(2,2))
  103. plot(sort(dat$dlem))
  104. plot(density(dat$dlem))
  105. qqnorm(dat$dlem)
  106. par(mfrow=c(1,1))
  107.  
  108. -------------------------
  109. #Frekvencija Leme (Kostic)
  110. -------------------------
  111.  
  112. par(mfrow=c(2,2))
  113. plot(sort(dat$flemk))
  114. plot(density(dat$flemk))
  115. qqnorm(dat$flemk)
  116. par(mfrow=c(1,1))
  117.  
  118. -------------------------
  119. #Frekvencija Leme (srWac)
  120. -------------------------
  121.  
  122. par(mfrow=c(2,2))
  123. plot(sort(dat$flemw))
  124. plot(density(dat$flemw))
  125. qqnorm(dat$flemw)
  126. par(mfrow=c(1,1))
  127.  
  128. -------------------------
  129. #Frekvencija Reci (srWac)
  130. -------------------------
  131.  
  132. par(mfrow=c(2,2))
  133. plot(sort(dat$fpov))
  134. plot(density(dat$fpov))
  135. qqnorm(dat$fpov)
  136. par(mfrow=c(1,1))
  137.  
  138. -------------------------
  139. #Frekvencija Sufiksa
  140. -------------------------
  141.  
  142. par(mfrow=c(2,2))
  143. plot(sort(dat$fsuf))
  144. plot(density(dat$fsuf))
  145. qqnorm(dat$fsuf)
  146. par(mfrow=c(1,1))
  147.  
  148. -------------------------
  149. #Duzina Sufiksa
  150. -------------------------
  151.  
  152. par(mfrow=c(2,2))
  153. plot(sort(dat$dsuf))
  154. plot(density(dat$dsuf))
  155. qqnorm(dat$dsuf)
  156. par(mfrow=c(1,1))
  157.  
  158. -------------------------
  159. #Produktivnost Sufiksa
  160. -------------------------
  161.  
  162. par(mfrow=c(2,2))
  163. plot(sort(dat$psuf))
  164. plot(density(dat$psuf))
  165. qqnorm(dat$psuf)
  166. par(mfrow=c(1,1))
  167.  
  168. __________________________________________________________
  169. ____________________________________________________________
  170.  
  171. # Vizuelna inspekcija slucajnih efekata
  172.  
  173. qqmath(~RT|Ispitanik,data=dat)
  174.  
  175. qqmath(~RT|TrialNumber,data=dat)
  176.  
  177. xylowess.fnc (RT~TrialOrder | Ispitanik, data=dat, ylab= "RT")
  178. ____________________________________________________________
  179. ____________________________________________________________
  180.  
  181. table(dat$FrekvencijaSufiksa)
  182.  
  183. 46 73 81 88 91 170 424 464 537 868 913 2140 2280
  184. 21 41 21 23 24 38 43 24 46 217 24 92 68
  185. 3001 3200 3764 3765 5345 6155 6585 23567
  186. 181 40 91 24 173 129 224 89
  187. _________________________________________________________
  188. ________________________________________________________________
  189.  
  190. # Kolinearnost medju prediktorima
  191.  
  192. C=cov(dat[,c("RT", "flemk","flemw", "fpov", "fsuf", "dlem","dsuf", "psuf")], y = NULL, use = "everything", method = c("pearson", "kendall", "spearman"))
  193. Cor=cov2cor(C)
  194. Cor
  195.  
  196. # RT flemk flemw fpov fsuf dlem
  197. #RT 1.0000000 -0.25270268 -0.340134744 -0.33576441 -0.2088414 0.11718241
  198. #flemk -0.2527027 1.00000000 0.737172303 0.81199371 0.4640850 -0.05401538
  199. #flemw -0.3401347 0.73717230 1.000000000 0.92810403 0.5037532 0.02816801
  200. #fpov -0.3357644 0.81199371 0.928104033 1.00000000 0.4884931 0.01208878
  201. #fsuf -0.2088414 0.46408501 0.503753210 0.48849305 1.0000000 -0.16956594
  202. #dlem 0.1171824 -0.05401538 0.028168006 0.01208878 -0.1695659 1.00000000
  203. #dsuf 0.1101909 -0.06037767 0.008320418 -0.05125295 -0.3250745 0.64594653
  204. #psuf -0.1763057 0.36969063 0.394772138 0.41680563 0.9190694 -0.11839922
  205. # dsuf psuf
  206. #RT 0.110190853 -0.1763057
  207. #flemk -0.060377670 0.3696906
  208. #flemw 0.008320418 0.3947721
  209. #fpov -0.051252949 0.4168056
  210. #fsuf -0.325074463 0.9190694
  211. #dlem 0.645946526 -0.1183992
  212. #dsuf 1.000000000 -0.3125367
  213. #psuf -0.312536700 1.0000000
  214.  
  215. collin.fnc(dat[,c("flemk","flemw", "fpov", "fsuf", "dlem","dsuf", "psuf")])$cnumber
  216.  
  217. #Naravno, prevelik zbog ovih frekvencija.
  218.  
  219. collin.fnc(dat[,c("flemk","fsuf", "dlem")])$cnumber
  220. #28.65964
  221.  
  222. _____________________________________________
  223. ____________________________________________________________
  224.  
  225. ################################################################# GLMER MODEL BROJ 1: FREKVENCIJA LEME (KOSTIC)
  226. ----------------------------
  227. ---------------------------------------------
  228. ---------------------------------------------------------
  229.  
  230. ########################## LMER KONACNI #######
  231. ################### SA NJE #################
  232.  
  233. glmer.dat6 <- glmer(Tacnost ~ poly(TrialOrder,2) + len.z + fsuf.z*NV + flemk.z + (1|Ispitanik) + (0+len.z|Ispitanik) +(1|Stimulus), data=dat, family="binomial")
  234. summary (glmer.dat6)
  235.  
  236. #Random effects:
  237. # Groups Name Variance Std.Dev.
  238. # Stimulus (Intercept) 2.01336 1.4189
  239. # Ispitanik len.z 0.02806 0.1675
  240. # Ispitanik.1 (Intercept) 0.22182 0.4710
  241. #Number of obs: 2023, groups: Stimulus, 88; Ispitanik, 46
  242. #
  243. #Fixed effects:
  244. # Estimate Std. Error z value Pr(>|z|)
  245. #(Intercept) 3.7379 0.3923 9.528 < 2e-16 ***
  246. #poly(TrialOrder, 2)1 9.6344 4.1242 2.336 0.01949 *
  247. #poly(TrialOrder, 2)2 0.7953 4.1678 0.191 0.84866
  248. #len.z 0.2887 0.2091 1.380 0.16748
  249. #fsuf.z 0.7625 0.2792 2.731 0.00632 **
  250. #NVvise 2.0246 1.0345 1.957 0.05033 .
  251. #flemk.z 1.0501 0.2493 4.213 2.52e-05 ***
  252. #fsuf.z:NVvise -4.0227 1.3319 -3.020 0.00253 **
  253.  
  254.  
  255. ---------------------
  256. dat$NV <- relevel(dat$NV, ref = "jedno")
  257.  
  258. glmer.dat6 <- glmer(Tacnost ~ poly(TrialOrder,2) + len.z + fsuf.z*NV + flemk.z + (1|Ispitanik) + (0+len.z|Ispitanik) +(1|Stimulus), data=dat, family="binomial")
  259. summary (glmer.dat6)
  260.  
  261. #Random effects:
  262. # Groups Name Variance Std.Dev.
  263. # Stimulus (Intercept) 2.01336 1.4189
  264. # Ispitanik len.z 0.02806 0.1675
  265. # Ispitanik.1 (Intercept) 0.22182 0.4710
  266. #Number of obs: 2023, groups: Stimulus, 88; Ispitanik, 46
  267. #
  268. #Fixed effects:
  269. # Estimate Std. Error z value Pr(>|z|)
  270. #(Intercept) 3.7379 0.3923 9.528 < 2e-16 ***
  271. #poly(TrialOrder, 2)1 9.6344 4.1242 2.336 0.01949 *
  272. #poly(TrialOrder, 2)2 0.7953 4.1678 0.191 0.84866
  273. #len.z 0.2887 0.2091 1.380 0.16748
  274. #fsuf.z 0.7625 0.2792 2.731 0.00632 **
  275. #NVvise 2.0246 1.0345 1.957 0.05033 .
  276. #flemk.z 1.0501 0.2493 4.213 2.52e-05 ***
  277. #fsuf.z:NVvise -4.0227 1.3319 -3.020 0.00253 **
  278.  
  279. ________________________________________________________
  280. __________________________________________________________________________________
  281. _________________________________________________________________________________________________________
  282.  
  283. ################################################################# MODEL BROJ 2: FREKVENCIJA LEME (SRWAC)
  284.  
  285. ----------------------------
  286. ---------------------------------------------
  287. ---------------------------------------------------------
  288.  
  289. ########################## TRENUTNO NAJBOLJI #######
  290. ################### SA NJE #################
  291.  
  292. glmer.dat3 <- glmer(Tacnost ~ poly(TrialOrder,2) + len.z + fsuf.z + flemw.z + NV + (1|Ispitanik) + (0+len.z|Ispitanik) +(1|Stimulus), data=dat, family="binomial")
  293. summary (glmer.dat3)
  294.  
  295. #Random effects:
  296. # Groups Name Variance Std.Dev.
  297. #Stimulus (Intercept) 1.69397 1.3015
  298. # Ispitanik len.z 0.03042 0.1744
  299. # Ispitanik.1 (Intercept) 0.22666 0.4761
  300. #Number of obs: 2023, groups: Stimulus, 88; Ispitanik, 46
  301. #
  302. #Fixed effects:
  303. # Estimate Std. Error z value Pr(>|z|)
  304. #(Intercept) 3.447492 0.349754 9.857 < 2e-16 ***
  305. #poly(TrialOrder, 2)1 10.408652 4.135174 2.517 0.0118 *
  306. #poly(TrialOrder, 2)2 1.126053 4.171276 0.270 0.7872
  307. #len.z 0.185551 0.194708 0.953 0.3406
  308. #fsuf.z 0.146900 0.252459 0.582 0.5606
  309. #flemw.z 1.366663 0.217854 6.273 3.53e-10 ***
  310. #NVvise -0.003134 0.512793 -0.006 0.9951
  311.  
  312. _____________________________________________
  313. ____________________________________________________________
  314. __________________________________________________________________________________
  315.  
  316. ################################################################# MODEL BROJ 3: FREKVENCIJA POVRSINSKA (SRWAC)
  317.  
  318. glmer.dat6 <- glmer(Tacnost ~ poly(TrialOrder,2) + len.z + fsuf.z + fpov.z + NV + (1|Ispitanik) + (0+len.z|Ispitanik) +(1|Stimulus), data=dat, family="binomial")
  319. summary (glmer.dat6)
  320.  
  321. #Random effects:
  322. # Groups Name Variance Std.Dev.
  323. # Stimulus (Intercept) 1.447e+00 1.20310
  324. # Ispitanik len.z 6.789e-05 0.00824
  325. # Ispitanik.1 (Intercept) 2.228e-01 0.47207
  326. #Number of obs: 2023, groups: Stimulus, 88; Ispitanik, 46
  327. #
  328. #Fixed effects:
  329. # Estimate Std. Error z value Pr(>|z|)
  330. #(Intercept) 3.42252 0.33428 10.238 < 2e-16 ***
  331. #poly(TrialOrder, 2)1 10.17160 4.11198 2.474 0.0134 *
  332. #poly(TrialOrder, 2)2 1.07198 4.13636 0.259 0.7955
  333. #len.z 0.21659 0.18052 1.200 0.2302
  334. #fsuf.z 0.17085 0.24129 0.708 0.4789
  335. #fpov.z 1.40548 0.20808 6.755 1.43e-11 ***
  336. #NVvise -0.02574 0.49405 -0.052 0.9585
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement
Public Pastes
Advertisement
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!