dashboard code.Rmd

---
title: "<Анализ данных проектов патформы Kickstarter>"
output: 
  flexdashboard::flex_dashboard:
    orientation: columns
    vertical_layout: fill
---

<style>
.navbar {
background-color:#6495ED;
border-color:paleturquoise;
}
.navbar-brand {
color:black!important;
}

</style>


Column {data-width=500}
-----------------------------------------------------------------------

### **Дерево зависимости статуса успешности проекта от некоторых переменных**

```{r}
library(flexdashboard)
library(stringr)
library(ggplot2)
library(dplyr)
library(data.table)
library(readr)
library(rpart)
library(rpart.plot)
library(coin)
library(stats)
set.seed(24601)
kickstarter = read_csv("~/shared/minor2_2020/1-Intro/hw2/kickstarter.csv")
goals = read_csv("~/shared/minor2_2020/1-Intro/hw2/usd_goals.csv")
kickstarter = full_join(kickstarter, goals, by = 'id')


set.seed(24601)
train = kickstarter %>% sample_frac(0.8) %>% dplyr::select(-name, -launched)
test = anti_join(kickstarter, train, by ='id') %>% dplyr::select(-name, -launched)


tree = rpart(state ~ ., data = train, control=rpart.control(cp=0.0207795735))
rpart.plot::rpart.plot(tree, cex = 0.8)
```

Column {data-width=500, data-height=50}
-----------------------------------------------------------------------

### **Точность модели на обучающей выборке**

```{r}

pred = predict(tree, type='class')
t = table(train$state, pred)
t0 = 100*(t[1,1] + t[2,2])/sum(t) 
t0_r = round(t0, digits = 1) 
t0_per = str_c(t0_r, "%")

valueBox(t0_per, icon = 'fa-graduation-cap', color = '#8FBC8F')

```

### **Точность модели на тестовой выборке**

```{r}

pred_test = predict(tree, type='class', newdata=test)
r = table(test$state, pred_test)
r0 = 100*(r[1,1] + r[2,2])/sum(r) 
r0_r = round(r0, digits = 1) 
r0_per = str_c(r0_r, "%")

valueBox(r0_per, icon = 'fa-file-alt', color = '#8FBC8F')

```

### Кроме usd_goal_real и backers, на успешность проекта также влияет category

```{r}


binar = ifelse(kickstarter$state == 'successful', 1, 0)
BIN = kickstarter %>% mutate(binar = binar)
BIN$binar = as.numeric(BIN$binar)
kickstarter1 = BIN %>% group_by(category) %>% summarise(total_num = length(state))

kickstarter2 = BIN %>% dplyr::select(category, binar) %>% group_by(category) %>% summarise(num_success = sum(binar))

joined = full_join(kickstarter1, kickstarter2, by = 'category') 
joined = joined %>% mutate(success_frequency = joined$num_success/joined$total_num) %>% arrange(-success_frequency) %>% filter(total_num>=50)

rund = round(100*joined$success_frequency, digits = 1)

jo = data.frame(Category = joined$category, Total_Num = joined$total_num, Num_Succeeded = joined$num_success, Success_Share = str_c(rund, '%'))

DT::datatable(jo, options = list(
bPaginate = TRUE
))
```