Email: abrahamzaweir@gmail.com
RPubs: https://rpubs.com/bram123

1 Regresi Linear Berganda

Dalam proyek ini saya memberikan anda dataset insurance.csv, informasi lanjut mengenai data ini dapat anda baca di Kaggle.

Tugas kalian adalah sebagai berikut:

Meringkas informasi penting yang terkadung data isurance.csv tersebut.
Memahami faktor-faktor apa yang mempengaruhi premi asuransi konsumen.
Menemukan model terbaik yang dapat memprediksi premi asuransi konsumen.

## Warning: package 'DT' was built under R version 3.6.3

## Warning: package 'Metrics' was built under R version 3.6.3

2 Insurance Member Behaviour

The purpose of this chapter is to get an insight of insurance member behaviour such as: age, sex, region, smoker.

2.1 Age of Insurance Member

The age distribution of insurance member is relatively the same, except 18 and 19 y’o members which has higher population (above 60). I also make a group of member’s age on table below.

##     Agecut total
## 1  [15,20]   166
## 2  (20,25]   140
## 3  (25,30]   138
## 4  (30,35]   130
## 5  (35,40]   127
## 6  (40,45]   137
## 7  (45,50]   144
## 8  (50,55]   140
## 9  (55,60]   125
## 10 (60,65]    91

2.2 Sex of Insurance Member

## Warning: The titlefont attribute is deprecated. Use title = list(font
## = ...) instead.

The gender of insurance member is almost the same (male = 676 people & female = 662 people).

2.3 Region of Insurance Member

## Warning: The titlefont attribute is deprecated. Use title = list(font
## = ...) instead.

The region where insurance members are living in is evenly distributed.

2.4 Insurance Member who Smoke

A lot of members are non smoker (79.5% or 1064 person) and the rest are smoker.

2.5 Number of Dependents

## Warning: The titlefont attribute is deprecated. Use title = list(font
## = ...) instead.

2.6 Body Mass Index of Insurance Member

Body mass index of members are normally distributed.

2.7 Charges of Insurance Member

The distribution of individual medical costs billed by health insurance has a positive skew.

3 What Variables Affect The Medical Cost?

I made this density analysis to find the variables which affect the medical costs. From this analysis (you need to see the density chart), I gain some insights:

Sex is not make an impact to medical costs insurance because male and female has a same distribution/density against charges (look at Chart 4.1).
Region is not make a big impact to medical costs insurance because 4 regions almost have a same distribution/density against charges (look at Chart 4.2).
Smoker and non-smoker affect the medical costs insurance because the distribution/density have a big different (look at Chart 4.3).
Total number of children/dependents have a same density against charges, except zero number of children. So, if you’re not have a child it will cause an effect to medical costs insurance (look at Chart 4.4).

According to those insights, I can conclude that Smoke, total dependents, age, and BMI are the variables to predict the medical costs insurance.

3.1 Charges Density Vs Sex

3.2 Charges Density Vs Region

3.3 Charges Density Vs Smoker

3.4 Charges Density Vs Children

4 Grouping The Equation

Since smokers, total dependents, and BMI made an effect to medical costs insurance, I’m grouping those factors into 8 category shown on sub-chapter below. Eventually, I only have age or and BMI as an input variable(s) to my prediction equation. The equation will shown in sub-chapter, so you need to check it.

Cat - 1 : Smoker, have no dependent, BMI under 30.

Cat - 2 : Smoker, have no dependent, BMI over 30.

Cat - 3 : Smoker, have dependents, BMI under 30.

Cat - 4 : Smoker, have dependents, BMI over 30.

Cat - 5 : Non-smoker, have no dependent, BMI under 30.

Cat - 6 : Non-smoker, have no dependent, BMI over 30.

Cat - 7 : Non-smoker, have dependents, BMI under 30.

Cat - 8 : Non-smoker, have dependents, BMI over 30.

4.1 Cat - 1

Smoker, have no dependent, BMI under 30.

Age vs Charges chart looks can be approached by using linear regression.

BMI vs Charges looks have a disordered correlation.

## 
## Call:
## lm(formula = charges ~ age, data = ins_smoker_nochild_under30)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -5740.7 -2372.9  -776.5   612.0 17119.6 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept) 11970.56    1454.78   8.228 5.55e-11 ***
## age           252.39      36.25   6.962 5.70e-09 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 4276 on 52 degrees of freedom
## Multiple R-squared:  0.4824, Adjusted R-squared:  0.4725 
## F-statistic: 48.46 on 1 and 52 DF,  p-value: 5.696e-09

## 
## Call:
## lm(formula = charges ~ age + bmi, data = ins_smoker_nochild_under30)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -2539.3 -1754.2 -1059.9  -203.1 15678.0 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  -956.74    5104.29  -0.187   0.8521    
## age           251.20      34.35   7.312 1.75e-09 ***
## bmi           505.18     192.06   2.630   0.0112 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 4051 on 51 degrees of freedom
## Multiple R-squared:  0.5442, Adjusted R-squared:  0.5264 
## F-statistic: 30.45 on 2 and 51 DF,  p-value: 1.985e-09

From those two summary, I decided to use linear regression with 2 variables (age and BMI) because it have a better R-Squared value.

So, the equation will be like:

-956.74 + (251.20 * AGE) + (505.18 * BMI)

4.2 Cat - 2

Smoker, have no dependent, BMI over 30.

Age vs Charges chart looks can be approached by using linear regression.

BMI vs Charges looks have a disordered correlation.

## 
## Call:
## lm(formula = charges ~ age, data = ins_smoker_nochild_over30)
## 
## Residuals:
##    Min     1Q Median     3Q    Max 
## -19722  -2239  -1235    786  19807 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept) 28983.17    1737.81  16.678  < 2e-16 ***
## age           306.74      43.38   7.071 2.05e-09 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 5361 on 59 degrees of freedom
## Multiple R-squared:  0.4587, Adjusted R-squared:  0.4495 
## F-statistic:    50 on 1 and 59 DF,  p-value: 2.05e-09

## 
## Call:
## lm(formula = charges ~ age + bmi, data = ins_smoker_nochild_over30)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -16667.0  -1505.1   -737.2     47.9  22684.4 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  8120.10    5338.78   1.521 0.133702    
## age           292.16      38.73   7.544 3.57e-10 ***
## bmi           614.01     150.40   4.082 0.000138 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 4766 on 58 degrees of freedom
## Multiple R-squared:  0.5795, Adjusted R-squared:  0.565 
## F-statistic: 39.97 on 2 and 58 DF,  p-value: 1.225e-11

From those two summary, I decided to use linear regression with 2 variables (age and BMI) because it have a better R-Squared value.

So, the equation will be like:

8120.10 + (292.16 * AGE) + (614.01 * BMI)

4.3 Cat - 3

Smoker, have dependents, BMI under 30.

Age vs Charges chart looks can be approached by using linear regression.

BMI vs Charges looks have a disordered correlation.

## 
## Call:
## lm(formula = charges ~ age, data = ins_smoker_child_under30)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -4651.4 -1487.2  -352.5   637.6 15865.3 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept) 10842.61    1332.43   8.137 7.77e-12 ***
## age           274.71      33.18   8.280 4.19e-12 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 3199 on 73 degrees of freedom
## Multiple R-squared:  0.4843, Adjusted R-squared:  0.4772 
## F-statistic: 68.56 on 1 and 73 DF,  p-value: 4.194e-12

## 
## Call:
## lm(formula = charges ~ age + bmi, data = ins_smoker_child_under30)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -2741.8 -1070.1  -608.8    -5.2 15619.8 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  2428.48    2769.81   0.877   0.3835    
## age           259.48      31.33   8.282 4.56e-12 ***
## bmi           359.27     105.64   3.401   0.0011 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 2989 on 72 degrees of freedom
## Multiple R-squared:  0.5557, Adjusted R-squared:  0.5433 
## F-statistic: 45.02 on 2 and 72 DF,  p-value: 2.075e-13

From those two summary, I decided to use linear regression with 2 variables (age and BMI) because it have a better R-Squared value.

So, the equation will be like:

2428.48 + (259.48 * AGE) + (359.27 * BMI)

4.4 Cat - 4

Smoker, have dependents, BMI over 30.

Age vs Charges chart looks can be approached by using linear regression.

BMI vs Charges looks have a disordered correlation.

## 
## Call:
## lm(formula = charges ~ age, data = ins_smoker_child_over30)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -4354.3 -2271.4  -859.3  1174.1 18445.4 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept) 32648.20    1361.98  23.971  < 2e-16 ***
## age           241.21      31.86   7.572 4.89e-11 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 3723 on 82 degrees of freedom
## Multiple R-squared:  0.4115, Adjusted R-squared:  0.4043 
## F-statistic: 57.34 on 1 and 82 DF,  p-value: 4.889e-11

## 
## Call:
## lm(formula = charges ~ age + bmi, data = ins_smoker_child_over30)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -2212.2 -1334.7  -653.6    40.7 17621.6 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept) 16021.03    3341.64   4.794 7.30e-06 ***
## age           253.72      27.69   9.162 3.81e-14 ***
## bmi           447.91      84.22   5.318 9.08e-07 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 3225 on 81 degrees of freedom
## Multiple R-squared:  0.5638, Adjusted R-squared:  0.553 
## F-statistic: 52.35 on 2 and 81 DF,  p-value: 2.553e-15

From those two summary, I decided to use linear regression with 2 variables (age and BMI) because it have a better R-Squared value.

So, the equation will be like:

16021.03 + (253.72 * AGE) + (447.91 * BMI)

4.5 Cat - 5

** Non-smoker, have no dependent, BMI under 30.**

Age vs Charges chart looks can be approached by using linear regression.

BMI vs Charges looks have a disordered correlation.

## 
## Call:
## lm(formula = charges ~ age, data = ins_nonsmoker_nochild_under30)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -2428.1 -1440.0  -886.5  -391.6 21672.8 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept) -3239.15     687.81  -4.709 4.43e-06 ***
## age           277.00      16.79  16.495  < 2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 3950 on 217 degrees of freedom
## Multiple R-squared:  0.5563, Adjusted R-squared:  0.5543 
## F-statistic: 272.1 on 1 and 217 DF,  p-value: < 2.2e-16

## 
## Call:
## lm(formula = charges ~ age + bmi, data = ins_nonsmoker_nochild_under30)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -2457.4 -1453.0  -852.6  -391.1 21715.4 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept) -3791.02    2191.60  -1.730   0.0851 .  
## age           276.56      16.91  16.354   <2e-16 ***
## bmi            22.40      84.44   0.265   0.7911    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 3958 on 216 degrees of freedom
## Multiple R-squared:  0.5565, Adjusted R-squared:  0.5524 
## F-statistic: 135.5 on 2 and 216 DF,  p-value: < 2.2e-16

From those two summary, I decided to use linear regression with 1 variable (age) because it have a better R-Squared value.

So, the equation will be like:

-3239.15 + (277.00 * AGE)

4.6 Cat - 6

** Non-smoker, have no dependent, BMI over 30.**

Age vs Charges chart looks can be approached by using linear regression.

BMI vs Charges looks have a disordered correlation.

## 
## Call:
## lm(formula = charges ~ age, data = ins_nonsmoker_nochild_over30)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -2595.7 -1666.9 -1096.7  -377.7 20412.6 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept) -2155.79     635.34  -3.393 0.000809 ***
## age           254.01      14.66  17.329  < 2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 3861 on 238 degrees of freedom
## Multiple R-squared:  0.5579, Adjusted R-squared:  0.556 
## F-statistic: 300.3 on 1 and 238 DF,  p-value: < 2.2e-16

## 
## Call:
## lm(formula = charges ~ age + bmi, data = ins_nonsmoker_nochild_over30)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -2811.9 -1632.1 -1099.0  -396.8 20324.2 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  -465.01    2319.97  -0.200    0.841    
## age           254.86      14.71  17.321   <2e-16 ***
## bmi           -48.90      64.53  -0.758    0.449    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 3864 on 237 degrees of freedom
## Multiple R-squared:  0.5589, Adjusted R-squared:  0.5552 
## F-statistic: 150.2 on 2 and 237 DF,  p-value: < 2.2e-16

From those two summary, I decided to use linear regression with 1 variable (age) because it have a better R-Squared value.

So, the equation will be like:

-2155.79 + (254.01 * AGE)

4.7 Cat - 7

** Non-smoker, have dependents, BMI under 30.**

Age vs Charges chart looks can be approached by using linear regression.

BMI vs Charges looks have a disordered correlation.

## 
## Call:
## lm(formula = charges ~ age, data = ins_nonsmoker_child_under30)
## 
## Residuals:
##    Min     1Q Median     3Q    Max 
##  -3050  -2080  -1578   -808  22413 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  -884.08    1029.59  -0.859    0.391    
## age           247.85      25.87   9.581   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 4975 on 281 degrees of freedom
## Multiple R-squared:  0.2462, Adjusted R-squared:  0.2436 
## F-statistic:  91.8 on 1 and 281 DF,  p-value: < 2.2e-16

## 
## Call:
## lm(formula = charges ~ age + bmi, data = ins_nonsmoker_child_under30)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -3255.8 -2178.9 -1528.4  -664.9 22367.6 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept) -2835.49    2564.04  -1.106    0.270    
## age           245.27      26.07   9.408   <2e-16 ***
## bmi            79.79      96.01   0.831    0.407    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 4978 on 280 degrees of freedom
## Multiple R-squared:  0.2481, Adjusted R-squared:  0.2427 
## F-statistic: 46.19 on 2 and 280 DF,  p-value: < 2.2e-16

From those two summary, I decided to use linear regression with 1 variable (age) because it have a better R-Squared value.

So, the equation will be like:

-884.08 + (247.85 * AGE)

4.8 Cat - 8

** Non-smoker, have dependents, BMI over 30.**

Age vs Charges chart looks can be approached by using linear regression.

BMI vs Charges looks have a disordered correlation.

## 
## Call:
## lm(formula = charges ~ age, data = ins_nonsmoker_child_over30)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -3138.6 -2283.0 -1647.7  -746.8 24235.0 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept) -2161.36    1041.35  -2.076   0.0387 *  
## age           282.54      24.23  11.660   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 5250 on 320 degrees of freedom
## Multiple R-squared:  0.2982, Adjusted R-squared:  0.296 
## F-statistic:   136 on 1 and 320 DF,  p-value: < 2.2e-16

## 
## Call:
## lm(formula = charges ~ age + bmi, data = ins_nonsmoker_child_over30)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -3320.7 -2278.4 -1636.0  -711.3 24252.0 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  -923.87    2648.40  -0.349    0.727    
## age           283.07      24.28  11.658   <2e-16 ***
## bmi           -35.83      70.50  -0.508    0.612    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 5256 on 319 degrees of freedom
## Multiple R-squared:  0.2988, Adjusted R-squared:  0.2944 
## F-statistic: 67.95 on 2 and 319 DF,  p-value: < 2.2e-16

From those two summary, I decided to use linear regression with 1 variable (age) because it have a better R-Squared value.

So, the equation will be like:

-2161.36 + (282.54 * AGE)

5 Charges Prediction

From grouping the equation, I made a function to predict the charges. The function is look like:

predict <- function(x){
               for(i in 1:nrow(x)){
               if(x[i,"smoker"] == "yes" && x[i,"children"] == 0 && x[i,"bmi"] < 30){
               x[i,"result"] = -956.74 + (251.20*x[i,"age"]) + (505.18*x[i,"bmi"])
               } else if(x[i,"smoker"] == "yes" && x[i,"children"] == 0 && x[i,"bmi"] >= 30) {
               x[i,"result"] = 8120.10 + (292.16*x[i,"age"]) + (614.01*x[i,"bmi"])
               } else if(x[i,"smoker"] == "yes" && x[i,"children"] > 0 && x[i,"bmi"] < 30){
               x[i,"result"] = 2428.48 + (259.48*x[i,"age"]) + (359.27*x[i,"bmi"])
               } else if(x[i,"smoker"] == "yes" && x[i,"children"] > 0 && x[i,"bmi"] >= 30){
               x[i,"result"] = 16021.03 + (253.72*x[i,"age"]) + (447.91*x[i,"bmi"])
               } else if(x[i,"smoker"] == "no" && x[i,"children"] == 0 && x[i,"bmi"] < 30){
               x[i,"result"] = -3239.15 + (277.00*x[i,"age"])
               } else if(x[i,"smoker"] == "no" && x[i,"children"] == 0 && x[i,"bmi"] >= 30){
               x[i,"result"] = -2155.79 + (254.01*x[i,"age"])
               } else if(x[i,"smoker"] == "no" && x[i,"children"] > 0 && x[i,"bmi"] < 30){
               x[i,"result"] = -884.08 + (247.85*x[i,"age"])
               } else {
               x[i,"result"] = -2161.36 + (282.54*x[i,"age"])
               }    
               }
          return(x)
          }

And the results of my prediction is shown on table below

The Root Mean Square Error of my prediction:

## [1] 4437.567

Mean error value of my prediction is about plus minus 4437.56

The Mean Absolute Percent Error of my prediction:

## [1] 0.2672371

Mean percentage error of my prediction is 26.7%

6 Study Cases

How much the insurance costs for this people?

Daniel, 27 years old, 34.21 BMI, have no dependent, non - smoker, and living in northeast.
Rani, 31 years old, 27.54 BMI, 3 dependents, smoker, and living in southeast.

Now, you can answer those question with this function:

predict_charge <- function(age, bmi, children, smoker){
    if(smoker == "yes" && children == 0 && bmi < 30){
        result = -956.74 + (251.20*age) + (505.18*bmi)
    } else if(smoker == "yes" && children == 0 && bmi >= 30) {
        result = 8120.10 + (292.16*age) + (614.01*bmi)
    } else if(smoker == "yes" && children > 0 && bmi < 30){
        result = 2428.48 + (259.48*age) + (359.27*bmi)
    } else if(smoker == "yes" && children > 0 && bmi >= 30){
        result = 16021.03 + (253.72*age) + (447.91*bmi)
    } else if(smoker == "no" && children == 0 && bmi < 30){
        result = -3239.15 + (277.00*age)
    } else if(smoker == "no" && children == 0 && bmi >= 30){
        result = -2155.79 + (254.01*age)
    } else if(smoker == "no" && children > 0 && bmi < 30){
        result = -884.08 + (247.85*age)
    } else {
        result = -2161.36 + (282.54*age)
    }
    return(result)
    }

The medical cost insurance for Daniel:

predict_charge(27, 34.21, 0, "no")

## [1] 4702.48

The medical cost insurance for Rani:

predict_charge(31, 27.54, 3, "yes")

## [1] 20366.66

7 ANOVA/MANOVA

Suatu perusahaan di Amerika Serikat ingin mempekerjakan seseorang dari luar Amerika Serikat untuk posisi teknis, mereka perlu mengajukan aplikasi ke pemerintah Amerika Serikat untuk mendapatkan kartu hijau atau visa bagi pelamar asing. Untuk menunjukkan ekuitas bagi karyawan AS dan non-AS, perusahaan perlu menyatakan seberapa banyak mereka bersedia membayar karyawan ketika mereka mengajukan permohonan visa atau kartu hijau. Sementara itu, mereka perlu memberikan jumlah rata-rata, yang disebut “prevailing wage” seorang karyawan dengan keterampilan dan latar belakang serupa biasanya dibayar untuk posisi yang sama.

Perbedaan antara upah yang dibayar dan upah yang berlaku dapat menunjukkan apakah perusahaan AS bersedia membayar lebih banyak gaji kepada karyawan non-AS. Gaji lebih banyak untuk calon karyawan asing akan menarik. Selain itu, perlu diperhatikan bahwa untuk area dan pekerjaan yang berbeda, gaji dapat menunjukkan perbedaan. Oleh karena itu perlu untuk mencari tahu hubungan antara gaji, area dan posisi dapat membantu karyawan non-AS untuk memilih pekerjaan di AS.

Berdasarkan klasifikasi VISA yang mereka miliki disimpulkan bahwa ada lima jenis yang berbeda: “green card”, “H-1B”, “H-1B1 Chile”, “H- 1B1 Singapore” dan “E-3 Australia”. Untuk projek ini, silahkan anda memilih kelas VISA “H-1B” untuk melakukan data mentah pelamar yang berpenduduk tetap tahun 2018 atau 2019. Kalian dapat mendwonload Data asli yang dikumpulkan oleh Kantor Sertifikasi Tenaga Kerja Asing Departemen Tenaga Kerja AS

Komputasi Statistika

UJian Akhir Semester

Abraham (20184920009)

20/12/2020

1 Regresi Linear Berganda

2 Insurance Member Behaviour

2.1 Age of Insurance Member

2.2 Sex of Insurance Member

2.3 Region of Insurance Member

2.4 Insurance Member who Smoke

2.5 Number of Dependents

2.6 Body Mass Index of Insurance Member

2.7 Charges of Insurance Member

3 What Variables Affect The Medical Cost?

3.1 Charges Density Vs Sex

3.2 Charges Density Vs Region

3.3 Charges Density Vs Smoker

3.4 Charges Density Vs Children

4 Grouping The Equation

4.1 Cat - 1

4.2 Cat - 2

4.3 Cat - 3

4.4 Cat - 4

4.5 Cat - 5

4.6 Cat - 6

4.7 Cat - 7

4.8 Cat - 8

5 Charges Prediction

6 Study Cases

7 ANOVA/MANOVA