tmp2.Rmd

## add benfold

```{r}
library(benford.analysis)
for (i 
     in 
        train_data_1 %>% 
        select(-short_name,-t,-tsi_real,-p) %>% 
        names
) {
    
    plot(benford(train_data_1[,i], number.of.digits=1))
     }
```

1. weedspeed 1
1. tsi 2
1. temperature 2

```{r}
var_is_numeric <- 
    train_data_1 %>% 
        select(-t,-p,-tsi_real) %>% 
        select_if(is.numeric) %>% 
        names
benfoldstart_time <- now()
benfoldmodel <- 
    prcomp_irlba(
        x = train_data_1 %>% 
            drop_data_underc %>% 
            select(var_is_numeric)
        ,n = 2
        ,center = T
        ,scale. = T
    )
benfoldend_time <- now()
benfoldend_time-benfoldstart_time
train_pca <- 
    predict(benfoldmodel
            ,newdata = 
              train_data_1 %>% 
              drop_data_underc %>% 
              select(var_is_numeric)
            )[[1]]
test_pca <- 
    predict(benfoldmodel
            ,newdata = 
              test_data_1 %>% 
              select(var_is_numeric)
            )[[1]]
```

```{r}
train_pca2 <- 
    predict(benfoldmodel
            ,newdata = 
              train_data_1 %>% 
              drop_data_underc %>% 
              select(var_is_numeric)
            )
test_pca2 <- 
    predict(benfoldmodel
            ,newdata = 
              test_data_1 %>% 
              select(var_is_numeric)
            )
```


```{r}
predict(benfoldmodel
    ,newdata = 
      train_data_1 %>% 
      drop_data_underc %>% 
      select(var_is_numeric)
    ) %>% 
    as.data.frame() %>% 
    ggplot(aes(PC1,PC2)) +
    geom_point(alpha=0.05)
```