RETO FINAL
Lorena Villarreal
2023-11-09
library(readxl)
variables <- read_excel("Base de datos RETO.xlsx",
sheet = "VARIABLES")na_count <- colSums(is.na(variables))
print(na_count)## edad_meses sexo escolaridad estrato violencia deprimido
## 0 0 8 0 0 440
## crecimiento castigo soledad soledadmenor
## 10 154 0 0variables <- na.omit(variables)
print(variables)## # A tibble: 393 × 10
## edad_meses sexo escolaridad estrato violencia deprimido crecimiento castigo
## <dbl> <chr> <dbl> <chr> <dbl> <dbl> <dbl> <chr>
## 1 33 Hombre 0 Rural … 0 0 4 No
## 2 25 Hombre 12 Urbano… 21 0 0 No
## 3 35 Hombre 15 Urbano… 21 0 1 No
## 4 26 Hombre 15 Urbano… 4 0 5 No
## 5 30 Hombre 15 Urbano… 0 0 12 Sí
## 6 35 Hombre 15 Urbano… 4 0 10 No
## 7 31 Hombre 15 Metrop… 10 0 0 No
## 8 31 Hombre 12 Metrop… 19 0 4 No
## 9 25 Hombre 12 Metrop… 0 52 4 No
## 10 32 Hombre 0 Metrop… 8 0 0 No
## # ℹ 383 more rows
## # ℹ 2 more variables: soledad <dbl>, soledadmenor <dbl>#Matriz de Correlación
cor_matrix <- cor(variables[, c("edad_meses", "escolaridad", "violencia", "deprimido", "crecimiento")])
print(cor_matrix)## edad_meses escolaridad violencia deprimido crecimiento
## edad_meses 1.00000000 -0.01809124 -0.02381185 0.03037982 0.09759053
## escolaridad -0.01809124 1.00000000 0.02459147 -0.07958463 0.16591401
## violencia -0.02381185 0.02459147 1.00000000 0.04739506 0.02985501
## deprimido 0.03037982 -0.07958463 0.04739506 1.00000000 -0.08988854
## crecimiento 0.09759053 0.16591401 0.02985501 -0.08988854 1.00000000interpret_correlation <- function(cor_value) {
if (cor_value > 0.8) {
return("Muy Fuerte")
} else if (cor_value > 0.6) {
return("Fuerte")
} else if (cor_value > 0.4) {
return("Moderado")
} else if (cor_value > 0.2) {
return("Debil")
} else {
return("Muy debil")
}
}
interpreted_matrix <- apply(cor_matrix, c(1, 2), interpret_correlation)
print(interpreted_matrix)## edad_meses escolaridad violencia deprimido crecimiento
## edad_meses "Muy Fuerte" "Muy debil" "Muy debil" "Muy debil" "Muy debil"
## escolaridad "Muy debil" "Muy Fuerte" "Muy debil" "Muy debil" "Muy debil"
## violencia "Muy debil" "Muy debil" "Muy Fuerte" "Muy debil" "Muy debil"
## deprimido "Muy debil" "Muy debil" "Muy debil" "Muy Fuerte" "Muy debil"
## crecimiento "Muy debil" "Muy debil" "Muy debil" "Muy debil" "Muy Fuerte"heatmap(cor_matrix,
annot = interpreted_matrix, # show the interpretations in each cell
cmap = "coolwarm", # color map
main = "Correlation Matrix with Interpretations")## Warning in plot.window(...): "annot" is not a graphical parameter## Warning in plot.window(...): "cmap" is not a graphical parameter## Warning in plot.xy(xy, type, ...): "annot" is not a graphical parameter## Warning in plot.xy(xy, type, ...): "cmap" is not a graphical parameter## Warning in title(...): "annot" is not a graphical parameter## Warning in title(...): "cmap" is not a graphical parameter
Visualización de Datos
#Distribución de edades por genero
ggplot(variables, aes(x = sexo, y = edad_meses, fill = sexo)) +
geom_violin() +
labs(title = "Distribución de Edades Por Género")
#Estrato
bar_social_status <- ggplot(variables, aes(x = factor(estrato))) +
geom_bar() +
labs(title = "Bar Chart: Distribución de Estrato", x = "Estrato", y = "Conteo")
print(bar_social_status)
#Violencia por edad
bar_age_violence <- ggplot(variables, aes(x = edad_meses, y = violencia)) +
geom_bar(stat = "identity", position = "dodge") +
labs(title = "Grouped Bar Chart: Violencia por Edad", x = "Edad en meses", y = "Violencia")
print(bar_age_violence)
#Horas de depresión
hist_depression <- ggplot(variables, aes(x = deprimido)) +
geom_histogram(binwidth = 5, fill = "skyblue", color = "white", alpha = 0.7) +
labs(title = "Histogram: Depression Hours", x = "Depression Hours", y = "Count")
print(hist_depression)
#Horas de depresión por genero
box_gender_depression <- ggplot(variables, aes(x = sexo, y = deprimido, fill = sexo)) +
geom_boxplot() +
labs(title = "Box Plot: Depression Hours by Gender", x = "Gender", y = "Depression Hours")
print(box_gender_depression)
#Depression por castigo (si/no)
bar_punishment_depression <- ggplot(variables, aes(x = castigo, fill = factor(deprimido))) +
geom_bar(position = "fill") +
labs(title = "Stacked Bar Chart: Depression por castigo (si/no)", x = "Punishment", y = "Proportion")
print(bar_punishment_depression)
Linear Regression - DEPRIMIDO
lm_model <- lm(deprimido ~ edad_meses + escolaridad + estrato + violencia + crecimiento + soledad + soledadmenor, data = variables)
summary(lm_model)##
## Call:
## lm(formula = deprimido ~ edad_meses + escolaridad + estrato +
## violencia + crecimiento + soledad + soledadmenor, data = variables)
##
## Residuals:
## Min 1Q Median 3Q Max
## -44.60 -20.46 -14.78 -5.98 357.26
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 25.0060 19.4750 1.284 0.200
## edad_meses 0.2585 0.3362 0.769 0.443
## escolaridad -1.3529 1.0021 -1.350 0.178
## estratoRural ( <25 -20.3519 27.6223 -0.737 0.462
## estratoUrbano ( 250 4.5690 7.6478 0.597 0.551
## violencia 0.3931 0.4359 0.902 0.368
## crecimiento -1.1766 0.7598 -1.549 0.122
## soledad -0.2351 1.9669 -0.120 0.905
## soledadmenor -0.0347 1.6315 -0.021 0.983
##
## Residual standard error: 66.03 on 384 degrees of freedom
## Multiple R-squared: 0.01904, Adjusted R-squared: -0.001396
## F-statistic: 0.9317 on 8 and 384 DF, p-value: 0.49summary_result <- summary(lm_model)
p_values <- summary_result$coefficients[, "Pr(>|t|)"]
alpha <- 0.05
for (i in seq_along(p_values)) {
if (p_values[i] < alpha) {
cat(names(p_values)[i], "is statistically significant (p-value <", alpha, ")\n")
} else {
cat(names(p_values)[i], "is not statistically significant (p-value >=", alpha, ")\n")
}
}## (Intercept) is not statistically significant (p-value >= 0.05 )
## edad_meses is not statistically significant (p-value >= 0.05 )
## escolaridad is not statistically significant (p-value >= 0.05 )
## estratoRural ( <25 is not statistically significant (p-value >= 0.05 )
## estratoUrbano ( 250 is not statistically significant (p-value >= 0.05 )
## violencia is not statistically significant (p-value >= 0.05 )
## crecimiento is not statistically significant (p-value >= 0.05 )
## soledad is not statistically significant (p-value >= 0.05 )
## soledadmenor is not statistically significant (p-value >= 0.05 )Linear Regression - VIOLENCIA
lm_model2 <- lm(violencia ~ edad_meses + escolaridad + estrato + deprimido + crecimiento + soledad + soledadmenor, data = variables)
summary(lm_model2)##
## Call:
## lm(formula = violencia ~ edad_meses + escolaridad + estrato +
## deprimido + crecimiento + soledad + soledadmenor, data = variables)
##
## Residuals:
## Min 1Q Median 3Q Max
## -10.401 -5.907 -3.057 3.208 17.371
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 6.373743 2.259065 2.821 0.00503 **
## edad_meses -0.021207 0.039336 -0.539 0.59011
## escolaridad 0.098444 0.117364 0.839 0.40211
## estratoRural ( <25 -1.082388 3.232057 -0.335 0.73789
## estratoUrbano ( 250 2.241118 0.887433 2.525 0.01196 *
## deprimido 0.005377 0.005961 0.902 0.36768
## crecimiento 0.078810 0.089036 0.885 0.37663
## soledad -0.081812 0.229989 -0.356 0.72224
## soledadmenor -0.570494 0.188559 -3.026 0.00265 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 7.722 on 384 degrees of freedom
## Multiple R-squared: 0.0477, Adjusted R-squared: 0.02787
## F-statistic: 2.405 on 8 and 384 DF, p-value: 0.01532summary_result2 <- summary(lm_model2)
p_values2 <- summary_result2$coefficients[, "Pr(>|t|)"]
alpha <- 0.05
for (i in seq_along(p_values2)) {
if (p_values2[i] < alpha) {
cat(names(p_values2)[i], "is statistically significant (p-value <", alpha, ")\n")
} else {
cat(names(p_values2)[i], "is not statistically significant (p-value >=", alpha, ")\n")
}
}## (Intercept) is statistically significant (p-value < 0.05 )
## edad_meses is not statistically significant (p-value >= 0.05 )
## escolaridad is not statistically significant (p-value >= 0.05 )
## estratoRural ( <25 is not statistically significant (p-value >= 0.05 )
## estratoUrbano ( 250 is statistically significant (p-value < 0.05 )
## deprimido is not statistically significant (p-value >= 0.05 )
## crecimiento is not statistically significant (p-value >= 0.05 )
## soledad is not statistically significant (p-value >= 0.05 )
## soledadmenor is statistically significant (p-value < 0.05 )Linear Regression - CRECIMIENTO
lm_model3 <- lm(crecimiento ~ edad_meses + escolaridad + estrato + deprimido + violencia + soledad + soledadmenor, data = variables)
summary(lm_model3)##
## Call:
## lm(formula = crecimiento ~ edad_meses + escolaridad + estrato +
## deprimido + violencia + soledad + soledadmenor, data = variables)
##
## Residuals:
## Min 1Q Median 3Q Max
## -6.216 -3.655 -1.153 3.684 19.579
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.670655 1.306359 0.513 0.60798
## edad_meses 0.045145 0.022413 2.014 0.04468 *
## escolaridad 0.207143 0.066425 3.118 0.00196 **
## estratoRural ( <25 0.511338 1.850660 0.276 0.78247
## estratoUrbano ( 250 -0.993561 0.509803 -1.949 0.05203 .
## deprimido -0.005275 0.003406 -1.549 0.12228
## violencia 0.025837 0.029189 0.885 0.37663
## soledad -0.181846 0.131379 -1.384 0.16712
## soledadmenor 0.060336 0.109199 0.553 0.58090
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 4.421 on 384 degrees of freedom
## Multiple R-squared: 0.05919, Adjusted R-squared: 0.03959
## F-statistic: 3.02 on 8 and 384 DF, p-value: 0.002662summary_result3 <- summary(lm_model3)
p_values3 <- summary_result3$coefficients[, "Pr(>|t|)"]
alpha <- 0.05
for (i in seq_along(p_values3)) {
if (p_values3[i] < alpha) {
cat(names(p_values3)[i], "is statistically significant (p-value <", alpha, ")\n")
} else {
cat(names(p_values3)[i], "is not statistically significant (p-value >=", alpha, ")\n")
}
}## (Intercept) is not statistically significant (p-value >= 0.05 )
## edad_meses is statistically significant (p-value < 0.05 )
## escolaridad is statistically significant (p-value < 0.05 )
## estratoRural ( <25 is not statistically significant (p-value >= 0.05 )
## estratoUrbano ( 250 is not statistically significant (p-value >= 0.05 )
## deprimido is not statistically significant (p-value >= 0.05 )
## violencia is not statistically significant (p-value >= 0.05 )
## soledad is not statistically significant (p-value >= 0.05 )
## soledadmenor is not statistically significant (p-value >= 0.05 )Multiple Regression - DEPRIMIDO
multi_model <- lm(deprimido ~ edad_meses + escolaridad + estrato + violencia + crecimiento + soledad + soledadmenor, data = variables)
summary(multi_model)##
## Call:
## lm(formula = deprimido ~ edad_meses + escolaridad + estrato +
## violencia + crecimiento + soledad + soledadmenor, data = variables)
##
## Residuals:
## Min 1Q Median 3Q Max
## -44.60 -20.46 -14.78 -5.98 357.26
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 25.0060 19.4750 1.284 0.200
## edad_meses 0.2585 0.3362 0.769 0.443
## escolaridad -1.3529 1.0021 -1.350 0.178
## estratoRural ( <25 -20.3519 27.6223 -0.737 0.462
## estratoUrbano ( 250 4.5690 7.6478 0.597 0.551
## violencia 0.3931 0.4359 0.902 0.368
## crecimiento -1.1766 0.7598 -1.549 0.122
## soledad -0.2351 1.9669 -0.120 0.905
## soledadmenor -0.0347 1.6315 -0.021 0.983
##
## Residual standard error: 66.03 on 384 degrees of freedom
## Multiple R-squared: 0.01904, Adjusted R-squared: -0.001396
## F-statistic: 0.9317 on 8 and 384 DF, p-value: 0.49summary_result_multi <- summary(multi_model)
p_values_multi <- summary_result_multi$coefficients[, "Pr(>|t|)"]
alpha_multi <- 0.05
for (i in seq_along(p_values_multi)) {
if (p_values_multi[i] < alpha_multi) {
cat(names(p_values_multi)[i], "is statistically significant (p-value <", alpha_multi, ")\n")
} else {
cat(names(p_values_multi)[i], "is not statistically significant (p-value >=", alpha_multi, ")\n")
}
}## (Intercept) is not statistically significant (p-value >= 0.05 )
## edad_meses is not statistically significant (p-value >= 0.05 )
## escolaridad is not statistically significant (p-value >= 0.05 )
## estratoRural ( <25 is not statistically significant (p-value >= 0.05 )
## estratoUrbano ( 250 is not statistically significant (p-value >= 0.05 )
## violencia is not statistically significant (p-value >= 0.05 )
## crecimiento is not statistically significant (p-value >= 0.05 )
## soledad is not statistically significant (p-value >= 0.05 )
## soledadmenor is not statistically significant (p-value >= 0.05 )Multiple Regression - VIOLENCIA
multi_model2 <- lm(violencia ~ edad_meses + escolaridad + estrato + deprimido + crecimiento + soledad + soledadmenor, data = variables)
summary(multi_model2)##
## Call:
## lm(formula = violencia ~ edad_meses + escolaridad + estrato +
## deprimido + crecimiento + soledad + soledadmenor, data = variables)
##
## Residuals:
## Min 1Q Median 3Q Max
## -10.401 -5.907 -3.057 3.208 17.371
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 6.373743 2.259065 2.821 0.00503 **
## edad_meses -0.021207 0.039336 -0.539 0.59011
## escolaridad 0.098444 0.117364 0.839 0.40211
## estratoRural ( <25 -1.082388 3.232057 -0.335 0.73789
## estratoUrbano ( 250 2.241118 0.887433 2.525 0.01196 *
## deprimido 0.005377 0.005961 0.902 0.36768
## crecimiento 0.078810 0.089036 0.885 0.37663
## soledad -0.081812 0.229989 -0.356 0.72224
## soledadmenor -0.570494 0.188559 -3.026 0.00265 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 7.722 on 384 degrees of freedom
## Multiple R-squared: 0.0477, Adjusted R-squared: 0.02787
## F-statistic: 2.405 on 8 and 384 DF, p-value: 0.01532summary_result_multi2 <- summary(multi_model2)
p_values_multi2 <- summary_result_multi2$coefficients[, "Pr(>|t|)"]
alpha_multi <- 0.05
for (i in seq_along(p_values_multi2)) {
if (p_values_multi2[i] < alpha_multi) {
cat(names(p_values_multi2)[i], "is statistically significant (p-value <", alpha_multi, ")\n")
} else {
cat(names(p_values_multi2)[i], "is not statistically significant (p-value >=", alpha_multi, ")\n")
}
}## (Intercept) is statistically significant (p-value < 0.05 )
## edad_meses is not statistically significant (p-value >= 0.05 )
## escolaridad is not statistically significant (p-value >= 0.05 )
## estratoRural ( <25 is not statistically significant (p-value >= 0.05 )
## estratoUrbano ( 250 is statistically significant (p-value < 0.05 )
## deprimido is not statistically significant (p-value >= 0.05 )
## crecimiento is not statistically significant (p-value >= 0.05 )
## soledad is not statistically significant (p-value >= 0.05 )
## soledadmenor is statistically significant (p-value < 0.05 )Multiple Regression - CRECIMIENTO
multi_model3 <- lm(crecimiento ~ edad_meses + escolaridad + estrato + deprimido + violencia + soledad + soledadmenor, data = variables)
summary(multi_model3)##
## Call:
## lm(formula = crecimiento ~ edad_meses + escolaridad + estrato +
## deprimido + violencia + soledad + soledadmenor, data = variables)
##
## Residuals:
## Min 1Q Median 3Q Max
## -6.216 -3.655 -1.153 3.684 19.579
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.670655 1.306359 0.513 0.60798
## edad_meses 0.045145 0.022413 2.014 0.04468 *
## escolaridad 0.207143 0.066425 3.118 0.00196 **
## estratoRural ( <25 0.511338 1.850660 0.276 0.78247
## estratoUrbano ( 250 -0.993561 0.509803 -1.949 0.05203 .
## deprimido -0.005275 0.003406 -1.549 0.12228
## violencia 0.025837 0.029189 0.885 0.37663
## soledad -0.181846 0.131379 -1.384 0.16712
## soledadmenor 0.060336 0.109199 0.553 0.58090
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 4.421 on 384 degrees of freedom
## Multiple R-squared: 0.05919, Adjusted R-squared: 0.03959
## F-statistic: 3.02 on 8 and 384 DF, p-value: 0.002662summary_result_multi3 <- summary(multi_model3)
p_values_multi3 <- summary_result_multi3$coefficients[, "Pr(>|t|)"]
alpha_multi <- 0.05
for (i in seq_along(p_values_multi3)) {
if (p_values_multi3[i] < alpha_multi) {
cat(names(p_values_multi3)[i], "is statistically significant (p-value <", alpha_multi, ")\n")
} else {
cat(names(p_values_multi3)[i], "is not statistically significant (p-value >=", alpha_multi, ")\n")
}
}## (Intercept) is not statistically significant (p-value >= 0.05 )
## edad_meses is statistically significant (p-value < 0.05 )
## escolaridad is statistically significant (p-value < 0.05 )
## estratoRural ( <25 is not statistically significant (p-value >= 0.05 )
## estratoUrbano ( 250 is not statistically significant (p-value >= 0.05 )
## deprimido is not statistically significant (p-value >= 0.05 )
## violencia is not statistically significant (p-value >= 0.05 )
## soledad is not statistically significant (p-value >= 0.05 )
## soledadmenor is not statistically significant (p-value >= 0.05 )ANOVA - DEPRIMIDO
anova_model <- aov(deprimido ~ edad_meses + escolaridad + estrato + violencia + crecimiento + soledad + soledadmenor, data = variables)
summary(anova_model)## Df Sum Sq Mean Sq F value Pr(>F)
## edad_meses 1 1575 1575 0.361 0.548
## escolaridad 1 10664 10664 2.446 0.119
## estrato 2 6543 3272 0.750 0.473
## violencia 1 3234 3234 0.742 0.390
## crecimiento 1 10404 10404 2.386 0.123
## soledad 1 72 72 0.017 0.897
## soledadmenor 1 2 2 0.000 0.983
## Residuals 384 1674185 4360p_value_overall <- summary(anova_model)[[1]]$`Pr(>F)`[1]
cat("Overall p-value:", p_value_overall, "\n")## Overall p-value: 0.5481478ANOVA - CRECIMIENTO
anova_model2 <- aov(crecimiento ~ edad_meses + escolaridad + estrato + violencia + deprimido + soledad + soledadmenor, data = variables)
summary(anova_model2)## Df Sum Sq Mean Sq F value Pr(>F)
## edad_meses 1 76 75.98 3.887 0.049371 *
## escolaridad 1 224 224.39 11.480 0.000776 ***
## estrato 2 73 36.66 1.875 0.154688
## violencia 1 13 13.30 0.680 0.409979
## deprimido 1 47 46.88 2.398 0.122276
## soledad 1 32 32.39 1.657 0.198776
## soledadmenor 1 6 5.97 0.305 0.580903
## Residuals 384 7506 19.55
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1p_value_overall2 <- summary(anova_model2)[[1]]$`Pr(>F)`[1]
cat("Overall p-value:", p_value_overall2, "\n")## Overall p-value: 0.04937086ANOVA - VIOLENCIA
anova_model3 <- aov(violencia ~ edad_meses + escolaridad + estrato + crecimiento + deprimido + soledad + soledadmenor, data = variables)
summary(anova_model3)## Df Sum Sq Mean Sq F value Pr(>F)
## edad_meses 1 14 13.6 0.229 0.63281
## escolaridad 1 14 14.0 0.235 0.62778
## estrato 2 399 199.3 3.342 0.03638 *
## crecimiento 1 41 41.3 0.693 0.40578
## deprimido 1 52 52.4 0.879 0.34899
## soledad 1 81 81.2 1.362 0.24400
## soledadmenor 1 546 545.8 9.154 0.00265 **
## Residuals 384 22896 59.6
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1p_value_overall3 <- summary(anova_model3)[[1]]$`Pr(>F)`[1]
cat("Overall p-value:", p_value_overall3, "\n")## Overall p-value: 0.6328086ANOVA - VIOLENCIA
anova_model <- aov(violencia ~ edad_meses + escolaridad + estrato + soledad + soledadmenor, data = variables)
summary(anova_model)## Df Sum Sq Mean Sq F value Pr(>F)
## edad_meses 1 14 13.6 0.229 0.63258
## escolaridad 1 14 14.0 0.236 0.62755
## estrato 2 399 199.3 3.347 0.03622 *
## soledad 1 90 90.1 1.513 0.21946
## soledadmenor 1 542 542.2 9.106 0.00272 **
## Residuals 386 22984 59.5
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Histograma
hist(variables$deprimido, breaks = 20, col = "skyblue", main = "Histograma de deprimido")
abline(v = median(variables$deprimido), col = "red", lty = 2)
Logistic Regression
table(variables$deprimido)##
## 0 12 52 365
## 339 13 28 13variables$binary_deprimido <- ifelse(variables$deprimido > 365, 1, 0)
table(variables$binary_deprimido)##
## 0
## 393logistic_model_deprimido <- glm(binary_deprimido ~ edad_meses + escolaridad + estrato + violencia + crecimiento + soledad + soledadmenor,
data = variables,
family = binomial)## Warning: glm.fit: algorithm did not convergesummary(logistic_model_deprimido)##
## Call:
## glm(formula = binary_deprimido ~ edad_meses + escolaridad + estrato +
## violencia + crecimiento + soledad + soledadmenor, family = binomial,
## data = variables)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.409e-06 -2.409e-06 -2.409e-06 -2.409e-06 -2.409e-06
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.657e+01 1.050e+05 0 1
## edad_meses -1.547e-15 1.813e+03 0 1
## escolaridad -1.180e-14 5.405e+03 0 1
## estratoRural ( <25 1.019e-12 1.490e+05 0 1
## estratoUrbano ( 250 4.711e-16 4.125e+04 0 1
## violencia -1.174e-15 2.351e+03 0 1
## crecimiento 1.461e-15 4.098e+03 0 1
## soledad 1.221e-15 1.061e+04 0 1
## soledadmenor -2.130e-15 8.799e+03 0 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 0.00e+00 on 392 degrees of freedom
## Residual deviance: 2.28e-09 on 384 degrees of freedom
## AIC: 18
##
## Number of Fisher Scoring iterations: 25#Interpretación de Logistic Regression
summary_result <- summary(logistic_model_deprimido)
# Interpretation of Null Deviance
cat("Null Deviance:\n")## Null Deviance:cat("The null deviance measures how well the model explains the response variable without any predictors.\n")## The null deviance measures how well the model explains the response variable without any predictors.cat("A smaller null deviance indicates that the model is better at explaining the outcome than a null model with no predictors.\n")## A smaller null deviance indicates that the model is better at explaining the outcome than a null model with no predictors.cat("Null Deviance:", summary_result$null.deviance, "on", summary_result$df.null, "degrees of freedom\n\n")## Null Deviance: 0 on 392 degrees of freedom# Interpretation of Residual Deviance
cat("Residual Deviance:\n")## Residual Deviance:cat("The residual deviance measures how well the model explains the response variable after including predictors.\n")## The residual deviance measures how well the model explains the response variable after including predictors.cat("A smaller residual deviance indicates a better fit of the model.\n")## A smaller residual deviance indicates a better fit of the model.cat("Residual Deviance:", summary_result$deviance, "on", summary_result$df.residual, "degrees of freedom\n\n")## Residual Deviance: 2.280022e-09 on 384 degrees of freedom# Interpretation of AIC
cat("AIC (Akaike Information Criterion):\n")## AIC (Akaike Information Criterion):cat("AIC is a measure of the model's goodness of fit, balancing the fit of the model with its complexity (number of parameters).\n")## AIC is a measure of the model's goodness of fit, balancing the fit of the model with its complexity (number of parameters).cat("A lower AIC indicates a better-fitting model.\n")## A lower AIC indicates a better-fitting model.cat("AIC:", summary_result$aic, "\n\n")## AIC: 18# Number of Fisher Scoring Iterations
cat("Number of Fisher Scoring Iterations:\n")## Number of Fisher Scoring Iterations:cat("The number of Fisher scoring iterations indicates how many iterations were performed by the optimization algorithm to find the maximum likelihood estimates of the model parameters.\n")## The number of Fisher scoring iterations indicates how many iterations were performed by the optimization algorithm to find the maximum likelihood estimates of the model parameters.cat("Number of Fisher Scoring Iterations:", summary_result$iter, "\n")## Number of Fisher Scoring Iterations: 25#Modelo Predictivo - DEPRIMIDO
print(paste("Modelo Predictivo - DEPRIMIDO"))## [1] "Modelo Predictivo - DEPRIMIDO"df_cleaned <- variables[, c("deprimido", "edad_meses", "escolaridad", "estrato", "soledad", "soledadmenor", "violencia", "crecimiento")]
df_cleaned <- df_cleaned[complete.cases(df_cleaned), ]
set.seed(123)
split_index <- sample(1:nrow(df_cleaned), 0.8 * nrow(df_cleaned))
train_data <- df_cleaned[split_index, ]
test_data <- df_cleaned[-split_index, ]
lm_model <- lm(deprimido ~ edad_meses + escolaridad + estrato + violencia + crecimiento, data = train_data)
predictions <- predict(lm_model, newdata = test_data)
mse <- mean((test_data$deprimido - predictions)^2)
r_squared <- summary(lm_model)$r.squared
adjusted_r_squared <- summary(lm_model)$adj.r.squared
mae <- mean(abs(test_data$deprimido - predictions))
rmse <- sqrt(mse)
print(paste("Model Evaluation Metrics:"))## [1] "Model Evaluation Metrics:"print(paste("Mean Squared Error (MSE): ", mse))## [1] "Mean Squared Error (MSE): 3515.18060247854"print(paste("Mean Absolute Error (MAE): ", mae))## [1] "Mean Absolute Error (MAE): 26.7148500999942"print(paste("Root Mean Squared Error (RMSE): ", rmse))## [1] "Root Mean Squared Error (RMSE): 59.288958520778"print(paste("R-squared: ", r_squared))## [1] "R-squared: 0.0230286612368626"print(paste("Adjusted R-squared: ", adjusted_r_squared))## [1] "Adjusted R-squared: 0.00393475885061223"if (mse == 0) {
print("The model perfectly predicts the 'deprimido' variable.")
} else {
print(paste("En promedio, las predicciones del modelo están equivocadas por aproximadamente", round(sqrt(mse), 2), "unidades en la escala de 'deprimido.'"))
}## [1] "En promedio, las predicciones del modelo están equivocadas por aproximadamente 59.29 unidades en la escala de 'deprimido.'"if (rmse < 10) {
cat("El RMSE es bajo, sugiriendo que el modelo tiene un buen ajuste.\n")
} else {
cat("El RMSE es alto, indicando que el modelo puede tener limitaciones.\n")
}## El RMSE es alto, indicando que el modelo puede tener limitaciones.if (r_squared > 0.5) {
cat("El R-squared es alto, indicando una buena proporción de varianza explicada por el modelo.\n")
} else {
cat("El R-squared is bajo, lo que sugiere que el modelo puede no capturar gran parte de la varianza en los datos.\n")
}## El R-squared is bajo, lo que sugiere que el modelo puede no capturar gran parte de la varianza en los datos.if (adjusted_r_squared > 0.5) {
cat("El Adjusted R-squared es alto, lo cual es favorable al considerar el número de predictores.\n")
} else {
cat("El Adjusted R-squared es bajo, sugiriendo que predictores adicionales pueden mejorar el modelo.\n")
}## El Adjusted R-squared es bajo, sugiriendo que predictores adicionales pueden mejorar el modelo.#Modelo Predictivo - VIOLENCIA
print(paste("Modelo Predictivo - VIOLENCIA"))## [1] "Modelo Predictivo - VIOLENCIA"df_cleaned <- variables[, c("deprimido", "edad_meses", "escolaridad", "estrato", "soledad", "soledadmenor", "violencia", "crecimiento")]
df_cleaned <- df_cleaned[complete.cases(df_cleaned), ]
set.seed(123)
split_index <- sample(1:nrow(df_cleaned), 0.8 * nrow(df_cleaned))
train_data <- df_cleaned[split_index, ]
test_data <- df_cleaned[-split_index, ]
lm_model2 <- lm(violencia ~ edad_meses + escolaridad + estrato + deprimido + crecimiento + soledad + soledadmenor, data = train_data)
predictions2 <- predict(lm_model2, newdata = test_data)
mse <- mean((test_data$violencia - predictions2)^2)
r_squared2 <- summary(lm_model2)$r.squared
adjusted_r_squared2 <- summary(lm_model2)$adj.r.squared
mae2 <- mean(abs(test_data$violencia - predictions2))
rmse2 <- sqrt(mae2)
print(paste("Model Evaluation Metrics:"))## [1] "Model Evaluation Metrics:"print(paste("Mean Squared Error (MSE): ", mse))## [1] "Mean Squared Error (MSE): 55.7045582737328"print(paste("Mean Absolute Error (MAE): ", mae))## [1] "Mean Absolute Error (MAE): 26.7148500999942"print(paste("Root Mean Squared Error (RMSE): ", rmse))## [1] "Root Mean Squared Error (RMSE): 59.288958520778"print(paste("R-squared: ", r_squared2))## [1] "R-squared: 0.0472922481856356"print(paste("Adjusted R-squared: ", adjusted_r_squared2))## [1] "Adjusted R-squared: 0.0223031924003408"if (mse == 0) {
print("The model perfectly predicts the 'violencia' variable.")
} else {
print(paste("En promedio, las predicciones del modelo están equivocadas por aproximadamente", round(sqrt(mse), 2), "unidades en la escala de 'violencia'"))
}## [1] "En promedio, las predicciones del modelo están equivocadas por aproximadamente 7.46 unidades en la escala de 'violencia'"if (rmse < 10) {
cat("El RMSE es bajo, sugiriendo que el modelo tiene un buen ajuste.\n")
} else {
cat("El RMSE es alto, indicando que el modelo puede tener limitaciones.\n")
}## El RMSE es alto, indicando que el modelo puede tener limitaciones.if (r_squared > 0.5) {
cat("El R-squared es alto, indicando una buena proporción de varianza explicada por el modelo.\n")
} else {
cat("El R-squared is bajo, lo que sugiere que el modelo puede no capturar gran parte de la varianza en los datos.\n")
}## El R-squared is bajo, lo que sugiere que el modelo puede no capturar gran parte de la varianza en los datos.if (adjusted_r_squared > 0.5) {
cat("El Adjusted R-squared es alto, lo cual es favorable al considerar el número de predictores.\n")
} else {
cat("El Adjusted R-squared es bajo, sugiriendo que predictores adicionales pueden mejorar el modelo.\n")
}## El Adjusted R-squared es bajo, sugiriendo que predictores adicionales pueden mejorar el modelo.#Modelo Predictivo - CRECIMIENTO
print(paste("Modelo Predictivo - CRECIMIENTO"))## [1] "Modelo Predictivo - CRECIMIENTO"df_cleaned <- variables[, c("deprimido", "edad_meses", "escolaridad", "estrato", "soledad", "soledadmenor", "violencia", "crecimiento")]
df_cleaned <- df_cleaned[complete.cases(df_cleaned), ]
set.seed(123)
split_index <- sample(1:nrow(df_cleaned), 0.8 * nrow(df_cleaned))
train_data <- df_cleaned[split_index, ]
test_data <- df_cleaned[-split_index, ]
lm_model3 <- lm(crecimiento ~ edad_meses + escolaridad + estrato + deprimido + violencia + soledad + soledadmenor, data = train_data)
predictions3 <- predict(lm_model3, newdata = test_data)
mse <- mean((test_data$crecimiento - predictions3)^2)
r_squared3 <- summary(lm_model3)$r.squared
adjusted_r_squared3 <- summary(lm_model3)$adj.r.squared
mae3 <- mean(abs(test_data$crecimiento - predictions3))
rmse3 <- sqrt(mae3)
print(paste("Model Evaluation Metrics:"))## [1] "Model Evaluation Metrics:"print(paste("Mean Squared Error (MSE): ", mse))## [1] "Mean Squared Error (MSE): 16.5057489199769"print(paste("Mean Absolute Error (MAE): ", mae))## [1] "Mean Absolute Error (MAE): 26.7148500999942"print(paste("Root Mean Squared Error (RMSE): ", rmse))## [1] "Root Mean Squared Error (RMSE): 59.288958520778"print(paste("R-squared: ", r_squared3))## [1] "R-squared: 0.0620773525638316"print(paste("Adjusted R-squared: ", adjusted_r_squared3))## [1] "Adjusted R-squared: 0.0374761027950142"if (mse == 0) {
print("The model perfectly predicts the 'violencia' variable.")
} else {
print(paste("En promedio, las predicciones del modelo están equivocadas por aproximadamente", round(sqrt(mse), 2), "unidades en la escala de 'violencia'"))
}## [1] "En promedio, las predicciones del modelo están equivocadas por aproximadamente 4.06 unidades en la escala de 'violencia'"if (rmse < 10) {
cat("El RMSE es bajo, sugiriendo que el modelo tiene un buen ajuste.\n")
} else {
cat("El RMSE es alto, indicando que el modelo puede tener limitaciones.\n")
}## El RMSE es alto, indicando que el modelo puede tener limitaciones.if (r_squared > 0.5) {
cat("El R-squared es alto, indicando una buena proporción de varianza explicada por el modelo.\n")
} else {
cat("El R-squared is bajo, lo que sugiere que el modelo puede no capturar gran parte de la varianza en los datos.\n")
}## El R-squared is bajo, lo que sugiere que el modelo puede no capturar gran parte de la varianza en los datos.if (adjusted_r_squared > 0.5) {
cat("El Adjusted R-squared es alto, lo cual es favorable al considerar el número de predictores.\n")
} else {
cat("El Adjusted R-squared es bajo, sugiriendo que predictores adicionales pueden mejorar el modelo.\n")
}## El Adjusted R-squared es bajo, sugiriendo que predictores adicionales pueden mejorar el modelo.