eval_pipe <- FALSE if(Sys.getenv("GLOBAL_EVAL") != "") eval_pipe <- Sys.getenv("GLOBAL_EVAL")
library(sparklyr) library(dplyr)
Create a simple estimator that transforms data and fits a model
Use the spark_lineitems
variable to create a new aggregation by order_id
. Summarize the total sales and number of items
```r
```
Assign the code to a new variable called orders
r
orders <-
Start a new code chunk, with calling ml_pipeline(sc)
r
ml_pipeline(sc)
Pipe the ml_pipeline()
code into a ft_dplyr_transfomer()
call. Use the orders
variable for its argument
```r
ml_pipeline(sc) %>%
```
Add an ft_binarizer()
step that determines if the total sale is above $50. Name the new variable above_50
r
ml_pipeline(sc) %>%
Using the ft_r_formula
, add a step that sets the model's formula to: above_50 ~ no_items
r
ml_pipeline(sc) %>%
Finalize the pipeline by adding a ml_logistic_regression()
step, no arguments are needed
r
ml_pipeline(sc) %>%
Assign the code to a new variable called orders_plan
r
orders_plan <- ml_pipeline(sc) %>%
Call orders_plan
to confirm that all of the steps are present
r
orders_plan
Execute the planned changes to obtain a new model
Use ml_fit()
to execute the changes in order_plan
using the spark_lineitems
data. Assign to a new variable called orders_fit
r
orders_fit <-
Call orders_fit
to see the print-out of the newly fitted model
r
orders_fit
Overview of how to use a fitted pipeline to run predictions
Use ml_transform()
in order to use the orders_fit
model to run predictions over spark_lineitems
r
orders_preds <- ml_transform(orders_fit, spark_lineitems)
With count()
, compare the results from above_50
against the predictions, the variable created by ml_transform()
is called prediction
```r
```
Overview of how to save the Estimator and the Transformer
Use ml_save()
to save order_plan
in a new folder called "saved_model"
```r
```
Navigate to the "saved_model" folder to inspect its contents
Use ml_save()
to save orders_fit
in a new folder called "saved_pipeline"
```r
```
Navigate to the "saved_pipeline" folder to inspect its contents
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.