blog post 7

library("caret")

Loading required package: ggplot2

Loading required package: lattice

library("corrplot")

corrplot 0.92 loaded

library("tidyverse")

── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
✔ dplyr     1.1.3     ✔ readr     2.1.4
✔ forcats   1.0.0     ✔ stringr   1.5.0
✔ lubridate 1.9.3     ✔ tibble    3.2.1
✔ purrr     1.0.2     ✔ tidyr     1.3.0

── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag()    masks stats::lag()
✖ purrr::lift()   masks caret::lift()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors

library(dplyr)
birth_sample<-read.csv("cleanedbirth.csv")
BirthStat <- read.csv("dataset/BirthStat.csv")
YI <- read.csv("dataset/Yearly income data.csv")
birthdata <-read.csv("dataset/cleanedbirth.csv")
combirth <- birthdata |>
  left_join(BirthStat, by = c("Year" = "Year"))
combirth <- combirth |>
  left_join(YI, by = c("Year" = "Year"))
birthweightmodel <- lm(dbwt ~ mager+ meduc+ m_ht_in+  bmi+ pwgt_r+ dwgt_r+  Married+ GDiabetes+ cig_before+ cig_during, data=birth_sample)
summary(birthweightmodel)

Call:
lm(formula = dbwt ~ mager + meduc + m_ht_in + bmi + pwgt_r + 
    dwgt_r + Married + GDiabetes + cig_before + cig_during, data = birth_sample)

Residuals:
    Min      1Q  Median      3Q     Max 
-3347.0  -289.1    35.3   347.1  7253.0 

Coefficients:
              Estimate Std. Error t value Pr(>|t|)    
(Intercept) 2130.88988   31.53166  67.579  < 2e-16 ***
mager         -4.84537    0.39108 -12.390  < 2e-16 ***
meduc          8.92433    1.30215   6.854 7.24e-12 ***
m_ht_in       17.59571    0.49910  35.255  < 2e-16 ***
bmi           -3.55587    0.36027  -9.870  < 2e-16 ***
pwgt_r         0.19442    0.03625   5.363 8.22e-08 ***
dwgt_r         0.39368    0.01918  20.523  < 2e-16 ***
Married       86.43417    4.22763  20.445  < 2e-16 ***
GDiabetes     -5.43742    7.37829  -0.737    0.461    
cig_before    67.20483   11.59933   5.794 6.90e-09 ***
cig_during  -259.91138    8.08793 -32.136  < 2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 596.4 on 119989 degrees of freedom
Multiple R-squared:  0.03344,   Adjusted R-squared:  0.03336 
F-statistic: 415.1 on 10 and 119989 DF,  p-value: < 2.2e-16

correlation_matrix <- 
  birth_sample|>
  select(-Race)|>
  cor()
corrplot(correlation_matrix, method = "circle", type = "upper", tl.col = "black", tl.srt = 45)

panel.cor <- function(x, y, digits = 2, prefix = "", cex.cor, ...) {
  usr <- par("usr")
  on.exit(par(usr))
  par(usr = c(0, 1, 0, 1))}
  
  
motheragemodel <- lm(mager ~ Year+meduc + m_ht_in + bmi + pwgt_r + dwgt_r + Married + GDiabetes + cig_before + cig_during+Year, data = birth_sample)
original_AIC <- AIC(motheragemodel)
commotheragemodel<- lm(mager ~ Year+meduc + m_ht_in + bmi + pwgt_r + dwgt_r + Married + GDiabetes + cig_before + cig_during+Year, data = combirth)
AIC(motheragemodel)

[1] 696234.4

motheragemodel <- lm(mager ~ Year+meduc+ m_ht_in+  bmi+ pwgt_r+ dwgt_r+  Married+ GDiabetes+ cig_before+ cig_during, data=birth_sample)
summary(motheragemodel)

Call:
lm(formula = mager ~ Year + meduc + m_ht_in + bmi + pwgt_r + 
    dwgt_r + Married + GDiabetes + cig_before + cig_during, data = birth_sample)

Residuals:
     Min       1Q   Median       3Q      Max 
-19.8342  -3.1093  -0.7844   2.3355  27.8346 

Coefficients:
              Estimate Std. Error t value Pr(>|t|)    
(Intercept) -1.006e+02  1.505e+01  -6.685 2.31e-11 ***
Year         5.761e-02  7.454e-03   7.729 1.09e-14 ***
meduc        1.565e+00  8.482e-03 184.520  < 2e-16 ***
m_ht_in      1.698e-02  3.683e-03   4.611 4.02e-06 ***
bmi          2.834e-02  2.659e-03  10.660  < 2e-16 ***
pwgt_r      -1.884e-03  2.676e-04  -7.039 1.95e-12 ***
dwgt_r       1.322e-03  1.415e-04   9.345  < 2e-16 ***
Married      2.844e+00  3.011e-02  94.477  < 2e-16 ***
GDiabetes    1.505e+00  5.430e-02  27.714  < 2e-16 ***
cig_before   2.830e-01  8.564e-02   3.304 0.000952 ***
cig_during   7.578e-01  5.966e-02  12.701  < 2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 4.402 on 119989 degrees of freedom
Multiple R-squared:  0.4179,    Adjusted R-squared:  0.4178 
F-statistic:  8614 on 10 and 119989 DF,  p-value: < 2.2e-16

AIC(motheragemodel)

[1] 696234.4

combirth$underage <- ifelse(birth_sample$mager < 18, 0, 1)
logistic_model1 <- glm(underage ~ Year+Number.of.births+Birth.rate.per.1.000.population.+Fertility.rate.births.per.1.000.women.aged.15.44. +Percent.born.low.birthweight+Percent.born.preterm+ Percent.unmarried+Mean.age.at.first.birth+PCI+meduc + m_ht_in + bmi + pwgt_r + dwgt_r + Married + GDiabetes + cig_before + cig_during, family = binomial(), data = combirth)
logistic_model2 <- glm(underage ~ Year+PCI+meduc + m_ht_in + bmi + pwgt_r + dwgt_r + Married + GDiabetes + cig_before + cig_during, 
                      family = binomial(), 
                      data = combirth)

logistic_model3 <- glm(underage ~ Year+meduc + m_ht_in + bmi + pwgt_r + dwgt_r + Married + GDiabetes + cig_before + cig_during, 
                      family = binomial(), 
                      data = combirth)
AIC(logistic_model1)

[1] 18959.16

AIC(logistic_model2)

[1] 18953.39

AIC(logistic_model3)

[1] 22206.89

summary(logistic_model3)

Call:
glm(formula = underage ~ Year + meduc + m_ht_in + bmi + pwgt_r + 
    dwgt_r + Married + GDiabetes + cig_before + cig_during, family = binomial(), 
    data = combirth)

Coefficients:
              Estimate Std. Error z value Pr(>|z|)    
(Intercept) -1.140e+02  2.104e+01  -5.417 6.06e-08 ***
Year         5.525e-02  1.042e-02   5.300 1.16e-07 ***
meduc        1.784e+00  2.549e-02  69.980  < 2e-16 ***
m_ht_in     -1.644e-02  5.659e-03  -2.905  0.00367 ** 
bmi          4.782e-02  5.233e-03   9.137  < 2e-16 ***
pwgt_r      -4.134e-03  5.000e-04  -8.269  < 2e-16 ***
dwgt_r       1.453e-03  2.560e-04   5.674 1.39e-08 ***
Married      3.336e+00  1.116e-01  29.890  < 2e-16 ***
GDiabetes    9.403e-01  1.289e-01   7.296 2.97e-13 ***
cig_before   6.958e-01  1.204e-01   5.777 7.60e-09 ***
cig_during   1.166e+00  7.998e-02  14.581  < 2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 37715  on 119999  degrees of freedom
Residual deviance: 22185  on 119989  degrees of freedom
AIC: 22207

Number of Fisher Scoring iterations: 10

set.seed(821)
train_indices <- sample(1:nrow(combirth), 0.7 * nrow(combirth))
train_data <- combirth[train_indices, ]
test_data <- combirth[-train_indices, ]
glmlogistic_model_train <- glm(underage ~ Year  + meduc + m_ht_in + bmi + pwgt_r + dwgt_r + Married + GDiabetes + cig_before + cig_during, 
                               family = binomial(), 
                               data = train_data)
predictions <- predict(glmlogistic_model_train, newdata = test_data, type = "response")
mse <- mean((test_data$underage - predictions)^2)
print(mse)

[1] 0.02787073

predicted_class <- ifelse(predictions > 0.5, 1, 0)
accuracy <- mean(predicted_class == test_data$underage)
print(accuracy)

[1] 0.9590556

Summary of fundings.

logistic_model2 has a slightly lower AIC compared to logistic_model1, suggesting a better balance between goodness of fit and model complexity. logistic_model3 has a higher AIC, indicating a less favorable model fit compared to the other two. Firstly, Our regression model(‘motheragemodel’) indicates a significant influence of mathernal age on birth outcomes. For every one-year increase in mathernal age, we observed a decreae in birth weight by about 4.85 grams.

Secondly, our model also highlights the positive impact of mathernal education(‘meduc’). With each additional year of education, there’s an estimated 8.92 grams increase in birth weight, and the p value is less than alpha, which shows the importance of education in maternal and child health.

Lastly, Smoking habits before and during pregnancy (cig_before, cig_during) shows mothers who smoked before pregnancy had babies weighing 67.20 grams more than non-smokers. In addition, for mother who smoking during pregnancy, their babies weighing 259.91 grams less. This result shows extreme negative effects of continued smoking during pregnancy.