r - Best practices using dplyr and timeSeries, ts?

Question

I have 14.5 years of budget data, contract IDs, project types, etc. with which I am trying to build an 18-month, time-series forecast. The data started out as individual payments on contract IDs by (non-continuous) date. Using Excel, I pivoted into total payments by month; later on I will include total active contracts in a month, composition of contract type, etc. There are a total of 3134 day-rows (out of 5296) on which payments where made - days on which no payments were made are not recorded in this data*.

The features I'm currently using are listed and structured as follows (not all features are below, just trying to get a model piped together using a linear t for now):

head(exp)
     Amount Day Month Year t
1  269909.4   5     7 2000 1
2  792078.6   6     7 2000 2
3  140065.5   7     7 2000 3
4  190553.2  11     7 2000 4
5  119208.6  12     7 2000 5
6 1068156.3  16     7 2000 6

> str(exp)
'data.frame':   3134 obs. of  5 variables:
 $ Amount: num  269909 792079 140066 190553 119209 ...
 $ Day   : int  5 6 7 11 12 16 17 21 26 28 ...
 $ Month : int  7 7 7 7 7 7 7 7 7 7 ...
 $ Year  : int  2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 ...
 $ t     : int  1 2 3 4 5 6 7 8 9 10 ...

I'm running into these problems / questions:

1. Dplyr is not at all liking the ts() objects I've used in my data.frame, so filtering and sorting by month/contract/contract type isn't working. What's the best approach here? I'm unsure about the pros/cons of using ts vs. timeSeries, especially as they relate to compatibility with other packages.

2. *Is this easier if I start with vectors of all 5296 days between 7/1/00 and 12/31/14, as well as a t <- 1:5296 and key these 3134 days of payments to that full list of days?

Answer 1

Waitting for answers