Rmd

dummies.Rmd

dummiesでダミー変数を作る
========================================================

```{r}
library(data.table)
library(dplyr)
library(dummies)

dat = fread("data.csv")

dat %>% head
```

```{r}
dat.dummy = dat %>% 
  mutate(hour = hour(datetime)) %>%  
  mutate(hour = as.factor(hour)) %>% 
  mutate(season = as.factor(season)) %>% 
  mutate(holiday = as.factor(holiday)) %>% 
  mutate(workingday = as.factor(workingday)) %>% 
  mutate(weather = as.factor(weather)) %>% 
  select(-datetime) %>% 
  dummy.data.frame()

library(glmnet)
dat.dummy.mat = 
  dat.dummy %>% 
  select(-hour23, -weather4, -holiday1, -workingday1, -season4) %>% 
  select(-casual, -registered, -count) %>% as.matrix 
dat.dummy.count = 
  dat.dummy %>% select(count) %>% unlist

dat.dummy.mat %>% head
glm_res = 
  glmnet(dat.dummy.mat, dat.dummy.count, family = "poisson", alpha = 0.1)
glm_res

predict(glm_res, dat.dummy.mat)
```