tmp.Rmd

# 数据报告 testing

```{r}
library(DataExplorer)
create_report(train_data_1)
```

直接用成绩pca一下。
不行!

# stacking testing

```{r}
library(here)
here('data') %>% 
    list.files('jiaxiang_prediction_xgboostbaseline_tweedie.csv') %>% 
    as_tibble() %>% 
    mutate(data = map(value,~fread(here('data',.)))) %>% 
    unnest() %>% 
    # filter(id == 6.5)
    spread(value,predicition) %>% 
    arrange(id) %>% 
    select(-id) %>% 
    prcomp_irlba(
        x = .
        ,n = 1
        ,center = F
        ,scale. = F
    ) %>% 
    predict() %>% 
    as_tibble() %>% 
    bind_cols(
        test_data_1 %>% 
        select(id)
    ) %>% 
    transmute(id,predicition=PC1) %>% 
    write_excel_csv(
        file.path(
            'data'
            ,paste(today() %>% str_remove_all('-') %>% str_sub(3,-1),'jiaxiang_prediction_xgboostbaseline_tweedie.csv',sep='_')
        )
    )
```