1-salary-linear-reg

简介

explore YearsExperience and Salary relationship

1. dataset: kaggle salary dataset

2. 数据类型需要做转换: to_ScienceType(d)=coerce(d,:YearsExperience=>Continuous,:Salary=>Continuous)

3. using MLJLinearModels.jl 🔗

1. load package

Code

    include("../utils.jl")
    import MLJ:fit!,fitted_params
    using GLMakie,MLJ,CSV,DataFrames

2. process data

load(csv)->dataframe ==>

Code

df=CSV.File("./data/salary_dataset.csv") |> DataFrame |> dropmissing;
first(df,5)

5×3 DataFrame

Row	Column1	YearsExperience	Salary
	Int64	Float64	Float64
1	0	1.2	39344.0
2	1	1.4	46206.0
3	2	1.6	37732.0
4	3	2.1	43526.0
5	4	2.3	39892.0

sciencetype ==>

Code

to_ScienceType(d)=coerce(d,:YearsExperience=>Continuous,:Salary=>Continuous)
new_df=to_ScienceType(df)
first(new_df,5)

5×3 DataFrame

Row	Column1	YearsExperience	Salary
	Int64	Float64	Float64
1	0	1.2	39344.0
2	1	1.4	46206.0
3	2	1.6	37732.0
4	3	2.1	43526.0
5	4	2.3	39892.0

MLJ table

Code

 X=MLJ.table(reshape(new_df[:,2],30,1))
 y=Vector(new_df[:,3])
 show(y)

[39344.0, 46206.0, 37732.0, 43526.0, 39892.0, 56643.0, 60151.0, 54446.0, 64446.0, 57190.0, 63219.0, 55795.0, 56958.0, 57082.0, 61112.0, 67939.0, 66030.0, 83089.0, 81364.0, 93941.0, 91739.0, 98274.0, 101303.0, 113813.0, 109432.0, 105583.0, 116970.0, 112636.0, 122392.0, 121873.0]

3. MLJ workflow

3.1 load model

  LinearRegressor = @load LinearRegressor pkg=MLJLinearModels
  model=LinearRegressor()
  mach = MLJ.fit!(machine(model,X,y))
  fp=MLJ.fitted_params(mach)  #学习的模型参数

[ Info: For silent loading, specify `verbosity=0`. 
[ Info: Training machine(LinearRegressor(fit_intercept = true, …), …).
┌ Info: Solver: MLJLinearModels.Analytical
│   iterative: Bool false
└   max_inner: Int64 200

import MLJLinearModels ✔

(coefs = [:x1 => 9449.962321455077],
 intercept = 24848.203966523164,)

3.2 build linear function

Code

    a=fp.coefs[1,1][2]
    b=fp.intercept
    line_func(t)=a*t+b

line_func (generic function with 1 method)

4. plot results

Code

xs=range(extrema(new_df[:,2])...,200)
fig=Figure()
ax=Axis(fig[1,1];xlabel="YearsExperience",ylabel="Salary")
lines!(ax,xs,line_func.(xs);label="fit-line",linewidth=3)
scatter!(ax,new_df[:,2],new_df[:,3];label="data",marker_style...)
axislegend(ax)
fig