pacman::p_load(tidyverse,
tidymodels,
janitor,
kableExtra,
GGally,
performance)Intro a Tidymodels
Carga de librerias
Datos
datos <- irisAnalisis Exploratorio de Datos
summary(datos) Sepal.Length Sepal.Width Petal.Length Petal.Width
Min. :4.300 Min. :2.000 Min. :1.000 Min. :0.100
1st Qu.:5.100 1st Qu.:2.800 1st Qu.:1.600 1st Qu.:0.300
Median :5.800 Median :3.000 Median :4.350 Median :1.300
Mean :5.843 Mean :3.057 Mean :3.758 Mean :1.199
3rd Qu.:6.400 3rd Qu.:3.300 3rd Qu.:5.100 3rd Qu.:1.800
Max. :7.900 Max. :4.400 Max. :6.900 Max. :2.500
Species
setosa :50
versicolor:50
virginica :50
ggpairs(datos)`stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
`stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
`stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
`stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
Particionamiento
set.seed(1234)
conf_part <- initial_split(datos, prop = 0.8)
datos_train <- training(conf_part)
datos_test <- testing(conf_part)
tabyl(datos_train$Species) datos_train$Species n percent
setosa 37 0.3083333
versicolor 42 0.3500000
virginica 41 0.3416667tabyl(datos_test$Species) datos_test$Species n percent
setosa 13 0.4333333
versicolor 8 0.2666667
virginica 9 0.3000000conf_part1 <- initial_split(datos, prop=0.8,
strata = Species)
datos_train1 <- training(conf_part1)
datos_test1 <- training(conf_part1)
tabyl(datos_train1$Species) datos_train1$Species n percent
setosa 40 0.3333333
versicolor 40 0.3333333
virginica 40 0.3333333Feature Engineering - recipes
datos_train1_sel <- datos_train1 |>
select(1:4)
datos_test1_sel <- datos_test1 |>
select(1:4)
receta <- recipe(Petal.Length ~ ., data = datos_train1_sel) |>
step_normalize(all_predictors()) |>
prep()
receta── Recipe ──────────────────────────────────────────────────────────────────────── Inputs Number of variables by roleoutcome: 1
predictor: 3── Training information Training data contained 120 data points and no incomplete rows.── Operations • Centering and scaling for: Sepal.Length and Sepal.Width, ... | TrainedAplicacion de la receta
datos_train1_sel_transf <- receta |>
juice()
mean(datos_train1_sel_transf$Sepal.Length)[1] -3.91608e-15sd(datos_train1_sel_transf$Sepal.Length)[1] 1datos_test1_sel_transf <- receta |>
bake(datos_test1_sel)Especificaciones estadisticas del modelo
Revisar los modelos existentes en el marco de la libreria parsnip
https://www.tidymodels.org/find/parsnip/
modelo_lineal <- linear_reg() |>
set_engine("lm") |>
set_mode("regression")Estamos ejecutando una tarea de regression.
Entrenamiento del modelo (fitting-ajuste)
ajuste_train_1 <- modelo_lineal |>
fit(Petal.Length ~ Sepal.Width, data = datos_train1_sel_transf)Vamos a hacer un parentesis del flujo ML dentro de tidymodels
ajuste_train_1_lm <- lm(Petal.Length ~ Sepal.Width, data = datos_train1_sel_transf)
summary(ajuste_train_1_lm)
Call:
lm(formula = Petal.Length ~ Sepal.Width, data = datos_train1_sel_transf)
Residuals:
Min 1Q Median 3Q Max
-3.7511 -1.4868 0.1675 1.2275 4.1561
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 3.7850 0.1501 25.222 < 2e-16 ***
Sepal.Width -0.7153 0.1507 -4.747 5.86e-06 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 1.644 on 118 degrees of freedom
Multiple R-squared: 0.1603, Adjusted R-squared: 0.1532
F-statistic: 22.53 on 1 and 118 DF, p-value: 5.862e-06ajuste_train_1 |>
pluck("fit") |>
summary()
Call:
stats::lm(formula = Petal.Length ~ Sepal.Width, data = data)
Residuals:
Min 1Q Median 3Q Max
-3.7511 -1.4868 0.1675 1.2275 4.1561
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 3.7850 0.1501 25.222 < 2e-16 ***
Sepal.Width -0.7153 0.1507 -4.747 5.86e-06 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 1.644 on 118 degrees of freedom
Multiple R-squared: 0.1603, Adjusted R-squared: 0.1532
F-statistic: 22.53 on 1 and 118 DF, p-value: 5.862e-06Aplicacion del modelo a la particion testing
ml_predicc_test <- ajuste_train_1 |> predict(datos_test1_sel_transf) |>
bind_cols(datos_test1_sel_transf)
ml_predicc_test_augment <- augment(ajuste_train_1, datos_test1_sel_transf)EVALUACION DEL MODELO (METRICAS DE AJUSTE)
ml_predicc_metricas <- ml_predicc_test |>
metrics(truth = Petal.Length, estimate = .pred)
ml_predicc_metricas# A tibble: 3 × 3
.metric .estimator .estimate
<chr> <chr> <dbl>
1 rmse standard 1.63
2 rsq standard 0.160
3 mae standard 1.40 Verificación de supuestos estadísticos
Metodo grafico
check_model(ajuste_train_1_lm)
Metodos pruebas estadisticas
check_autocorrelation(ajuste_train_1_lm)Warning: Autocorrelated residuals detected (p < .001).check_heteroscedasticity(ajuste_train_1_lm)OK: Error variance appears to be homoscedastic (p = 0.092).check_normality(ajuste_train_1_lm)Warning: Non-normality of residuals detected (p = 0.029).autocorrelacion que hay dependencia en los residuos se busca que los residuos sean de variaza constante
Modelo regresión lineal multiple TALLER
Particionamiento 2
set.seed(1234)
part2 <- initial_split(datos, prop = 0.8)
datos_train2 <- training(conf_part)
datos_test2 <- testing(conf_part)
tabyl(datos_train2$Species) datos_train2$Species n percent
setosa 37 0.3083333
versicolor 42 0.3500000
virginica 41 0.3416667tabyl(datos_test2$Species) datos_test2$Species n percent
setosa 13 0.4333333
versicolor 8 0.2666667
virginica 9 0.3000000part1 <- initial_split(datos, prop=0.8,
strata = Species)
datos_train2 <- training(part1)
datos_test2 <- training(part1)
tabyl(datos_train2$Species) datos_train2$Species n percent
setosa 40 0.3333333
versicolor 40 0.3333333
virginica 40 0.3333333Feature Engineering - recipes 2
datos_train2_sel <- datos_train2 |>
select(1:4)
datos_test2_sel <- datos_test2 |>
select(1:4)
receta2 <- recipe(Petal.Length ~ ., data = datos_train2_sel) |>
step_normalize(all_predictors()) |>
prep()
receta── Recipe ──────────────────────────────────────────────────────────────────────── Inputs Number of variables by roleoutcome: 1
predictor: 3── Training information Training data contained 120 data points and no incomplete rows.── Operations • Centering and scaling for: Sepal.Length and Sepal.Width, ... | TrainedAplicacion de la receta 2
datos_train2_sel_transf <- receta |>
juice()
mean(datos_train2_sel_transf$Sepal.Length)[1] -3.91608e-15sd(datos_train2_sel_transf$Sepal.Length)[1] 1datos_test2_sel_transf <- receta |>
bake(datos_test2_sel)Especificaciones estadisticas del modelo2
Revisar los modelos existentes en el marco de la libreria parsnip
modelo_lineal2 <- linear_reg() |>
set_engine("lm") |>
set_mode("regression")Estamos ejecutando una tarea de regression.
Entrenamiento del modelo (fitting-ajuste) 2
ajuste_train_2 <- modelo_lineal2 |>
fit(Petal.Length ~ Petal.Width, data = datos_train1_sel_transf)Vamos a hacer un parentesis del flujo ML dentro de tidymodels
ajuste_train_2_lm <- lm(Sepal.Length ~ Sepal.Width + Petal.Length + Petal.Width, data = datos_train2_sel_transf)
summary(ajuste_train_2_lm)
Call:
lm(formula = Sepal.Length ~ Sepal.Width + Petal.Length + Petal.Width,
data = datos_train2_sel_transf)
Residuals:
Min 1Q Median 3Q Max
-0.90328 -0.26336 0.01991 0.22737 0.98184
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -3.15850 0.28572 -11.054 < 2e-16 ***
Sepal.Width 0.33482 0.03836 8.728 2.23e-14 ***
Petal.Length 0.83448 0.07495 11.134 < 2e-16 ***
Petal.Width -0.49795 0.13000 -3.830 0.000208 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 0.374 on 116 degrees of freedom
Multiple R-squared: 0.8636, Adjusted R-squared: 0.8601
F-statistic: 244.9 on 3 and 116 DF, p-value: < 2.2e-16ajuste_train_2 |>
pluck("fit") |>
summary()
Call:
stats::lm(formula = Petal.Length ~ Petal.Width, data = data)
Residuals:
Min 1Q Median 3Q Max
-1.15612 -0.32195 0.00475 0.25925 1.38425
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 3.78500 0.04428 85.47 <2e-16 ***
Petal.Width 1.71989 0.04447 38.68 <2e-16 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 0.4851 on 118 degrees of freedom
Multiple R-squared: 0.9269, Adjusted R-squared: 0.9263
F-statistic: 1496 on 1 and 118 DF, p-value: < 2.2e-16Aplicacion del modelo a la particion testing 2
ml_predicc_test2 <- ajuste_train_2 |> predict(datos_test2_sel_transf) |>
bind_cols(datos_test2_sel_transf)
ml_predicc_test_augment <- augment(ajuste_train_2, datos_test2_sel_transf)EVALUACION DEL MODELO (METRICAS DE AJUSTE) 2
ml_predicc_metricas2 <- ml_predicc_test2 |>
metrics(truth = Petal.Length, estimate = .pred)
ml_predicc_metricas# A tibble: 3 × 3
.metric .estimator .estimate
<chr> <chr> <dbl>
1 rmse standard 1.63
2 rsq standard 0.160
3 mae standard 1.40 Metodo grafico2
check_model(ajuste_train_2_lm)
Metodos pruebas estadisticas 2
check_autocorrelation(ajuste_train_2_lm)OK: Residuals appear to be independent and not autocorrelated (p = 0.998).check_heteroscedasticity(ajuste_train_2_lm)Warning: Heteroscedasticity (non-constant error variance) detected (p = 0.041).check_normality(ajuste_train_2_lm)OK: residuals appear as normally distributed (p = 0.865).