BTC forecasting models R

btc.Rmd

---
title: "R Notebook"
output: html_notebook
---


```{r}
library(forecast)
library(curl)
library(rjson)
yesterday = Sys.Date()-1

start = "2013-08-01"

url <- paste0("https://api.coindesk.com/v1/bpi/historical/close.json?start=", start, "&end=", yesterday)

curl_download(url, destfile = "btc.json")
json_data <- fromJSON(file = "btc.json")

raw.data <- as.data.frame(t(as.data.frame(json_data$bpi)))$V1

ts <- ts(raw.data, start =1, frequency = 12)
autoplot(ts, xlab = "Months Since 2013-08-01", ylab = "Bitcoin Price Index (USD)")

```
```{r}
ndiffs(ts)
ts <- diff(ts)
ndiffs(ts)
```


```{r}
model1 <- ses(ts)
model2 <- holt(ts)
model3 <- hw(ts)
model4 <- auto.arima(ts)
pred4 <- predict(model4, 7)
pred4

autoplot(model1)
autoplot(model2)
autoplot(model3)
#ts.plot(ts, pred4)
```


Sorry about the lack of docs. Here's a book instead: https://otexts.com/fpp2/.