using DataFrames
using FreqTables

auto = readtable("auto.csv");

showcols(auto)

num_obs  = size(auto,1)

num_vars = size(auto,2)

auto[:price]

auto[2]

auto[:,[:price,:mpg]]

head(auto,4)

tail(auto,4)

auto[[1;2;4;15],[:headroom,:trunk]]

auto[(auto[:,:mpg].<20),[:headroom,:trunk]]

describe(auto)

colwise(mean,auto)

round(colwise(mean,auto[:,2:end]),3)

countmap(auto[:foreign])

by(auto,:foreign,nrow)

freqtable(auto, :rep78, :foreign, subset=!isna(auto[:rep78]))

auto1 = auto[!isna(auto[:,:rep78]), :];

showcols(auto1)

auto1 = auto1[setdiff(names(auto1), [:price,:mpg])];

showcols(autod1)

delete!(auto1,[:weight,:length]);

showcols(auto1)

auto2 = auto[:,[:make,:mpg,:displacement,:gear_ratio]];

showcols(auto2)

rename!(auto2,[:make,:displacement],[:make_name,:CCs]);

showcols(auto2)

auto2[:mpg_same] = auto2[:mpg];

showcols(auto2)

auto2[:mpgSquared] = map(temp -> temp.^2, auto[:mpg]);

head(auto2[:,[:mpg,:mpgSquared]])

auto2[:dummy_var] = map((tempx,tempy) -> (tempx.<20) & (tempy.<3), auto2[:mpg], auto2[:gear_ratio]);

showcols(auto2)

auto2 = auto2[:,[2;3;4;1;5:end]];

showcols(auto2)

sort!(auto2,cols=[:mpg,:make_name],rev=[false,true]);

head(auto2,4)

reshape1 = DataFrame(id = 1:3, sex = [0;1;0], 
                     inc1980 = [5000;2000;3000],
                     inc1981 = [5500;2200;2000],
                     inc1982 = [6000;4400;1000])

longform1A = stack(reshape1, [:inc1980, :inc1981, :inc1982], [:id, :sex])

sort!(longform1A, cols = [:id, :variable])

wideform1A = unstack(longform1A, :variable, :value)

name = DataFrame(ID = [1, 2, 3, 4, 5, 6], Name = ["John", "Jane", "Mark", "Ann", "Vlad", "Maria"])

jobs = DataFrame(ID = [1, 2, 3, 4, 5, 6], Job = ["Lawyer", "Doctor", "Mechanic", "Doctor", "Judge", "Pilot"])

siblings = DataFrame(ID = [1, 1, 2, 3, 5, 5, 5, 6],
           Sibling = ["Eric", "Ryan", "Jennifer", "Heather", "Carl", "Dmitri", "Andrei", "Pedro"])

mergedNameJobs = join(name,jobs, on = :ID, kind = :inner)

mergedNameSibsInner = join(name,siblings, on = :ID, kind = :inner)

mergedNameSibsOuter = join(name,siblings, on = :ID, kind = :outer)

mergedNameSibsLeft = join(name,siblings, on = :ID, kind = :left)

mergedNameSibsOuter = join(name,siblings, on = :ID, kind = :right)

mergedNameSibsSemi = join(name,siblings, on = :ID, kind = :semi)

mergedNameSibsAnti = join(name,siblings, on = :ID, kind = :anti)

mergedNameSibsCross = join(name,siblings, kind = :cross)

reshape2 = DataFrame(id = 1:3, sex = [0;1;0], inc1980 = [5000;2000;3000], 
                     inc1981 = [5500;2200;2000],inc1982 = [6000;4400;1000],
                    ue1980 = [0;1;0], ue1981 = [1;0;0], ue1982 = [0;0;1])

longform2 = stack(reshape2, [:inc1980, :inc1981, :inc1982, :ue1980, :ue1981, :ue1982],
                 [:id, :sex])