Problem in chapter 5 : Master Spark with R

[FluffyPanda0310]

Hi,
In chapter 5 : Master Spark with R, when I try to run these code, it worked

okc_train %>%
ft_string_indexer(input_col = "sex", output_col = "sex_indexed") %>%
ft_string_indexer(input_col = "drinks", output_col = "drinks_indexed") %>%
ft_string_indexer(input_col = "drugs", output_col = "drugs_indexed") %>%
select(age, sex_indexed, drinks_indexed, drugs_indexed, essay_length) %>%
ft_one_hot_encoder_estimator(
input_cols = c("sex_indexed", "drinks_indexed", "drugs_indexed"),
output_cols = c("sex_encoded", "drinks_encoded", "drugs_encoded")
)

But the original one :

pipeline <- ml_pipeline(sc) %>%
ft_string_indexer(input_col = "sex", output_col = "sex_indexed") %>%
ft_string_indexer(input_col = "drinks", output_col = "drinks_indexed") %>%
ft_string_indexer(input_col = "drugs", output_col = "drugs_indexed") %>%
ft_one_hot_encoder_estimator(
input_cols = c("sex_indexed", "drinks_indexed", "drugs_indexed"),
output_cols = c("sex_encoded", "drinks_encoded", "drugs_encoded"))
pipeline_model <- ml_fit(pipeline, okc_train)

gave out the error :

Error in as.character(call[[1]]) :
cannot coerce type 'closure' to vector of type 'character'

I don't know where the error comes from, since the vectors are all in numeric not in character.

[problemxsolutions]

> traceback()
17: spark_require_version(spark_connection(jobj), "3.0.0")
16: (function (jobj) 
    {
        spark_require_version(spark_connection(jobj), "3.0.0")
        new_ml_transformer(jobj, category_size = invoke(jobj, "categorySize"), 
            class = "ml_one_hot_encoder_model")
    })(jobj = <environment>)
15: do.call(ml_get_constructor(jobj), list(jobj = jobj))
14: FUN(X[[i]], ...)
13: lapply(stages, ml_call_constructor)
12: (function (jobj, ..., class = character()) 
    {
        stages <- tryCatch({
            jobj %>% invoke("stages")
        }, error = function(e) {
            NULL
        })
        if (!rlang::is_na(stages)) {
            stages <- lapply(stages, ml_call_constructor)
        }
        new_ml_transformer(jobj, stages = stages, stage_uids = if (rlang::is_null(stages)) {
            NULL
        }
        else {
            sapply(stages, function(x) {
                x$uid
            })
        }, ..., class = c(class, "ml_pipeline_model"))
    })(jobj = <environment>)
11: do.call(ml_get_constructor(jobj), list(jobj = jobj))
10: ml_call_constructor(invoke(jobj, "bestModel"))
9: structure(list(uid = invoke(jobj, "uid"), param_map = ml_get_param_map(jobj), 
       ..., .jobj = jobj), class = c(class, "ml_pipeline_stage"))
8: new_ml_pipeline_stage(jobj, ..., class = c(class, "ml_transformer"))
7: new_ml_transformer(jobj, estimator = invoke(jobj, "getEstimator") %>% 
       ml_call_constructor(), evaluator = invoke(jobj, "getEvaluator") %>% 
       ml_call_constructor(), estimator_param_maps = ml_get_estimator_param_maps(jobj), 
       best_model = ml_call_constructor(invoke(jobj, "bestModel")), 
       ..., class = c(class, "ml_tuning_model"))
6: new_ml_tuning_model(jobj, num_folds = invoke(jobj, "getNumFolds"), 
       metric_name = metric_name, avg_metrics = avg_metrics, avg_metrics_df = ml_get_estimator_param_maps(jobj) %>% 
           param_maps_to_df() %>% dplyr::mutate(`:=`(!!metric_name, 
           avg_metrics)) %>% dplyr::select(!!metric_name, dplyr::everything()), 
       sub_models = possibly_null(~jobj %>% invoke("subModels") %>% 
           purrr::map(~purrr::map(.x, ml_call_constructor))), class = "ml_cross_validator_model")
5: (function (jobj) 
   {
       avg_metrics <- invoke(jobj, "avgMetrics")
       metric_name <- jobj %>% invoke("getEvaluator") %>% invoke("getMetricName") %>% 
           rlang::sym()
       new_ml_tuning_model(jobj, num_folds = invoke(jobj, "getNumFolds"), 
           metric_name = metric_name, avg_metrics = avg_metrics, 
           avg_metrics_df = ml_get_estimator_param_maps(jobj) %>% 
               param_maps_to_df() %>% dplyr::mutate(`:=`(!!metric_name, 
               avg_metrics)) %>% dplyr::select(!!metric_name, dplyr::everything()), 
           sub_models = possibly_null(~jobj %>% invoke("subModels") %>% 
               purrr::map(~purrr::map(.x, ml_call_constructor))), 
           class = "ml_cross_validator_model")
   })(jobj = <environment>)
4: do.call(ml_get_constructor(jobj), list(jobj = jobj))
3: ml_call_constructor(.)
2: spark_jobj(x) %>% invoke("fit", spark_dataframe(dataset)) %>% 
       ml_call_constructor()
1: ml_fit(x = cv, dataset = okc_train)