aaaa

[vinciusb]

[vinciusb]

O código faz o seguinte, ele filtra os dados pela função filter, depois ele seleciona apenas algumas colunas, fiz uma regrassão simples para achá-las, depois pesquisa mais se tem mais alguma que você quer incluir, mas faz sentido se você olhar. O mutate altera as colunas para não ter vírgula

stack_overflow = read_csv("C:/Users/Guilherme Namem/Repositorios/StackOverflow/survey_results_public.csv") v = stack_overflow %>% select(ConvertedCompYearly) %>% na.omit() %>% pull(ConvertedCompYearly) SUP = quantile(v, .75) + 1.5 IQR(v) INF = quantile(v, .25) - 1.5 IQR(v) d = stack_overflow %>% filter(MainBranch == 'I am a developer by profession', ConvertedCompYearly >= INF, ConvertedCompYearly <= SUP) %>% select(Age, Employment, RemoteWork, EdLevel, YearsCode, YearsCodePro, DevType, OrgSize, Country, CompTotal, WorkExp, ConvertedCompYearly) %>% na.omit() %>% mutate(Age =factor(Age), Employment=factor(Employment), RemoteWork = factor(RemoteWork), EdLevel =factor(EdLevel), YearsCode=as.integer(ifelse(YearsCodePro == 'Less than 1 year', 0, ifelse(YearsCodePro == 'More than 50 years', 50, YearsCodePro))), YearsCodePro=factor(YearsCodePro), DevType = factor(DevType), OrgSize = factor(OrgSize), Country=factor(Country), WorkExp=factor(WorkExp), ConvertedCompYearly = log10(ConvertedCompYearly))

d = stack_overflow %>% filter( MainBranch == 'I am a developer by profession', ConvertedCompYearly >= INF, ConvertedCompYearly <= SUP, Age != 'Under 18 years old', Age != 'Prefer not to say', Employment %in% c( 'Employed, full-time;Employed, part-time', 'Employed, part-time', "Independent contractor, freelancer, or self-employed;Employed, part-time") ) %>% select( Age, Employment, RemoteWork, EdLevel, YearsCode, YearsCodePro, DevType, OrgSize, Country, CompTotal, WorkExp, ConvertedCompYearly) %>% mutate( Age =factor(gsub(',','',Age)), Employment=factor(gsub(',','',Employment) ), RemoteWork = factor(gsub(',','',RemoteWork)), EdLevel =factor(gsub(',','',EdLevel)), YearsCode=as.integer(ifelse(YearsCode == 'Less than 1 year', 0, ifelse(YearsCode == 'More than 50 years', 50, YearsCode))), YearsCodePro=as.integer(ifelse(YearsCodePro == 'Less than 1 year', 0, ifelse(YearsCodePro == 'More than 50 years', 50, YearsCodePro))), DevType = factor(gsub(',','',DevType)), OrgSize = factor(gsub(',','',OrgSize) ), Country=factor(gsub(',','',Country)), ConvertedCompYearly=log10(ConvertedCompYearly))