时间序列建模

Question

I have a dependent variable (target) which looks like this: This is daily data of exchange rate of EUR/USD.

ExchangeDate | ExchangeRate   
2012-01-01   |   0.772484   
2012-01-02   |   0.773471   
2012-01-03   |   0.766388   
2012-01-04   |   0.772803   
2012-01-05   |   0.781781 

I have some independent variables which are like unemployment rate, GDP of countries etc. These are either monthly or quarterly data. For example, unemployment data for Austria looks like this.

YrQtr   |   UnempRate          
2012-Q1 |   4.553893   
2012-Q2 |   4.915041   
2012-Q3 |   5.204023   
2012-Q4 |   5.042323   
2013-Q1 |   5.470267  

How do I convert this to a daily time series so that the dependent and independent variables have the granularity? In this case, how do I convert this quarterly data into daily?

My idea is to ultimately use the ARIMA model.

Answer 1

This issue is resolved. i used cubic interpolation for the conversion. Here is my code.

## Using Cubic interpolation, converting quarterly data to monthly
# Quarter in year-quarter format
currAcctBal$quarter = as.yearqtr(currAcctBal$YrQtr,format="%Y-Q%q")

# Quarter in year-month-day format
currAcctBal$qvar = as.Date(currAcctBal$quarter)

# Creating a monthly sequence for the quarterly range
monthly = seq(currAcctBal$qvar[1], tail(currAcctBal$qvar, 1), by="month")

# Adding 2 more months to the list
monthly = c(monthly, as.Date("2015-11-01"))
monthly = c(monthly, as.Date("2015-12-01"))

##### Cubic interpolation using spline()
### Getting variable of interest for Austria
subAus = subset(currAcctBal, select = c("qvar", "currAcctBalRate.AUT"))
currAcctBalAus = data.frame(qvar=monthly, AcctBal=spline(subAus, 
method="fmm", xout=monthly)$y)