تحلیل داده‌های زلزله با R – بخش دوم

suppressPackageStartupMessages(library(ggplot2))
suppressPackageStartupMessages(library(dplyr))
suppressPackageStartupMessages(library(Hmisc))
suppressPackageStartupMessages(library(lubridate))
suppressPackageStartupMessages(library(vcd))
suppressPackageStartupMessages(library(vcdExtra))
suppressPackageStartupMessages(library(gmodels))

packages <- c("ggplot2", "dplyr", "Hmisc", "lubridate", "vcd", "vcdExtra", "gmodels")
version <- lapply(packages, packageVersion)
version_c <- do.call(c, version)
data.frame(packages=packages, version = as.character(version_c))

if ("all_week.csv" %in% dir(".") == FALSE) {
  url <- "https://earthquake.usgs.gov/earthquakes/feed/v1.0/summary/all_month.csv"
  download.file(url = url, destfile = "all_week.csv")
}
quakes <- read.csv("all_month.csv", header=TRUE, sep=',', stringsAsFactors = FALSE)

quakes$time <- ymd_hms(quakes$time)
quakes$updated <- ymd_hms(quakes$updated)
quakes$magType <- as.factor(quakes$magType)
quakes$net <- as.factor(quakes$net)
quakes$type <- as.factor(quakes$type)
quakes$status <- as.factor(quakes$status)
quakes$locationSource <- as.factor(quakes$locationSource)
quakes$magSource <- as.factor(quakes$magSource)

(factor_vars <- names(which(sapply(quakes, class) == "factor")))
## [1] "magType"        "net"            "type"           "status"        
## [5] "locationSource" "magSource"
length(factor_vars)
## [1] 6

describe(quakes[,factor_vars])
## quakes[, factor_vars] 
## 
##  6  Variables      8407  Observations
## ---------------------------------------------------------------------------
## magType 
##        n  missing distinct 
##     8407        0       10 
##                                                                       
## Value               mb mb_lg    md    mh    ml   mun    mw   mwr   mww
## Frequency      2   604    47  2423    14  5203     2     4    19    89
## Proportion 0.000 0.072 0.006 0.288 0.002 0.619 0.000 0.000 0.002 0.011
## ---------------------------------------------------------------------------
## net 
##        n  missing distinct 
##     8407        0       14 
## 
## ak (2469, 0.294), ci (1344, 0.160), hv (253, 0.030), ismpkansas (8,
## 0.001), ld (4, 0.000), mb (157, 0.019), nc (1435, 0.171), nm (28, 0.003),
## nn (604, 0.072), pr (427, 0.051), se (15, 0.002), us (897, 0.107), uu
## (588, 0.070), uw (178, 0.021)
## ---------------------------------------------------------------------------
## type 
##        n  missing distinct 
##     8407        0        7 
## 
## chemical explosion (2, 0.000), earthquake (8232, 0.979), explosion (58,
## 0.007), ice quake (16, 0.002), other event (3, 0.000), quarry blast (95,
## 0.011), rock burst (1, 0.000)
## ---------------------------------------------------------------------------
## status 
##        n  missing distinct 
##     8407        0        2 
##                               
## Value      automatic  reviewed
## Frequency       1691      6716
## Proportion     0.201     0.799
## ---------------------------------------------------------------------------
## locationSource 
##        n  missing distinct 
##     8407        0       15 
##                                                                       
## Value         ak    ci    hv  ismp    ld    mb    nc    nm    nn    ok
## Frequency   2470  1344   253     8     4   157  1435    28   604     6
## Proportion 0.294 0.160 0.030 0.001 0.000 0.019 0.171 0.003 0.072 0.001
##                                         
## Value         pr    se    us    uu    uw
## Frequency    427    15   890   588   178
## Proportion 0.051 0.002 0.106 0.070 0.021
## ---------------------------------------------------------------------------
## magSource 
##        n  missing distinct 
##     8407        0       15 
##                                                                       
## Value         ak    ci    hv  ismp    ld    mb    nc    nm    nn    ok
## Frequency   2480  1344   253     8     4   157  1435    28   604     5
## Proportion 0.295 0.160 0.030 0.001 0.000 0.019 0.171 0.003 0.072 0.001
##                                         
## Value         pr    se    us    uu    uw
## Frequency    427    15   881   588   178
## Proportion 0.051 0.002 0.105 0.070 0.021
## ---------------------------------------------------------------------------

quakes$magType[quakes$magType == ""] <- NA