LSTM_temp.Rmd

---
title: "MCM_2022"
author: "MCM2022"
date: "2/17/2022"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

```{r}
library(dplyr)
library(ggplot2)
library(ggthemes)
library(lubridate)

theme_set(theme_bw())

set.seed(123)
```

```{r}
# Set colors
DB <- rgb(29/255, 59/255, 83/255)
B <- rgb(69/255, 147/255, 182/255)
O <- rgb(225/255, 116/255, 56/255)
```

```{r}
# Load the data & clean the data
bitcoin_data <- read.csv("data/BCHAIN-MKPRU.csv") %>% 
  mutate(Date = as.Date(Date, "%m/%d/%y")) %>% 
  na.omit()

gold_data <- read.csv("data/LBMA-GOLD.csv") %>% 
  mutate(Date = as.Date(Date, "%m/%d/%y")) %>% 
  na.omit()

bitcoinLSTM <- read.csv("DataPrediction/Output/LSTM/bitcoin_LSTM.csv")
bitcoinARIMA <- readRDS("DataPrediction/Output/ARIMA/bitcoin_ARIMA.RData")

bitcoinLSTM <- bitcoinLSTM %>% 
  mutate(Type = "predict-LSTM",
         Value = value) %>% 
  select(Date, Value, Type)
  
bitcoinARIMA <- bitcoin_data %>% 
  filter(Date > "2020-09-09") %>% 
  mutate(Value = bitcoinARIMA,
         Type = "predict-ARIMA")

bitcoin_final <- bitcoin_data %>% 
  mutate(Type = "original-raw") %>% 
  rbind(bitcoinLSTM) %>% 
  rbind(bitcoinARIMA)
```

```{r}
bitcoin_final %>% 
  ggplot(aes(x = Date,
             y = Value,
             color = Type)) +
  geom_line() +
  scale_color_manual(values = c(O, B, DB))

bitcoin_final %>%
  filter(Date >= "2020-09-09") %>% 
  ggplot(aes(x = Date,
             y = Value,
             color = Type)) +
  geom_line() +
  scale_color_manual(values = c(O, B, DB))
```

```{r}
MSE_ARIMA <- bitcoin_data %>% 
  filter(Date >= "2020-09-10") %>% 
  mutate(Raw = Value) %>%
  select(-Date, -Value) %>% 
  cbind(bitcoinARIMA) %>% 
  mutate(dif = (Raw -Value)^2) %>% 
  summarise(MSE = sum(dif) / n()) %>% 
  pull()

MSE_ARIMA
```