Merge pull request #2 from edgararuiz/wip

Wip

.gitignore

@@ -5,3 +5,4 @@
 logs
 derby.log
 sense.txt
+tuberia

sparklyr.Rmd

@@ -125,7 +125,6 @@ muestra_vuelos$entrenar
 modelo <- muestra_vuelos$entrenar %>%
   ml_logistic_regression(tarde ~.)
 
-
 ```
 
 ## Visualizaciones
@@ -170,6 +169,78 @@ vuelos %>%
 ```
 
 
+# Pipelines (Tuberias)
+
+```{r}
+entrenar <- muestra_vuelos$entrenar %>%
+  mutate(
+    arr_delay = ifelse(arr_delay == "NaN", 0, arr_delay)
+  ) %>%
+  select(
+    month,
+    sched_dep_time,
+    arr_delay,
+    distance
+  ) %>%
+  mutate_all(as.numeric)
+```
+
+
+```{r}
+tuberia_vuelos <- ml_pipeline(sc) %>%
+  ft_dplyr_transformer(
+    tbl = entrenar
+  ) %>% 
+  ft_binarizer(
+    input.col = "arr_delay",
+    output.col = "tarde",
+    threshold = 15
+  ) %>%
+  ft_bucketizer(
+    input.col = "sched_dep_time",
+    output.col = "horas",
+    splits = c(400, 800, 1200, 1600, 2000, 2400)
+  )  %>%
+  ft_r_formula(tarde ~ horas + distance + arr_delay) %>% 
+  ml_logistic_regression()
+
+tuberia_vuelos
+
+```
+
+```{r}
+modelo_nuevo <- ml_fit(
+  tuberia_vuelos, 
+  muestra_vuelos$entrenar
+  )
+
+modelo_nuevo
+```
+
+```{r}
+predicciones <- ml_transform(
+  x = modelo_nuevo,
+  dataset = muestra_vuelos$examinar
+)
+
+predicciones
+```
+
+```{r}
+predicciones%>%
+  group_by(tarde, prediction) %>%
+  tally()
+```
+
+```{r}
+ml_save(tuberia_vuelos, "tuberia", overwrite = TRUE)
+
+dir("tuberia")
+```
+
+
+# Analysis de texto
+
 ```{r}
 library(janeaustenr)
 ```