I have a data.frame consisting of 2 variables with each 2.5 million obs.
str(values)
data.frame': 2529905 obs. of 2 variables:
$ Date : Factor w/ 498 levels "1977-11","1978-06",..: 108 60 12 108 58 108 132 188 51 60 ...
$ Value: num 223000 171528 110269 426000 172436 ...
> head(values)
Date Value
1 2003-01 223000.0
2 1999-01 171528.0
3 1992-01 110268.6
4 2003-01 426000.0
5 1998-11 172436.5
6 2003-01 334000.0
I wanted to make a data.frame with the median per date:
library(plyr)
medianperdate = ddply(values, .(Date), summarize, median_value = median(Value))
> str(medianperdate)
'data.frame': 498 obs. of 2 variables:
$ Date : Factor w/ 498 levels "1977-11","1978-06",..: 1 2 3 4 5 6 7 8 9 10 ...
$ median_value: num 106638 84948 85084 75725 88487 ...
> head(medianperdate)
Date median_value
1 1977-11 106638.35
2 1978-06 84947.65
3 1985-07 85083.79
4 1986-05 75724.58
5 1986-11 88487.14
6 1986-12 98697.20
But what I want, is an extra column which counts the observations per month (eg. 2003-01, the data used would be object "values"
And another extra column where I define which class house it is:
a : < 200 000
b : < 300 000 & > 200 000
c : < 300 000 & > 2000000
I will continuetrying this but because I am already stuck for a couple of hours I will appreciate help very much!!
If it is not clear, what I can understand. The following testdataframe presents how I would like my dataframe to look like
> testdf
Year MedianValue HouseClass #Observations
1 1999-1 200000 B 501
2 1999-2 150000 A 664
3 1999-3 250000 C 555
Like my answer to your previous question 0
library(data.table)
dt <- data.table(df)
dt2 <- dt[,list(
medianvalue = median(value),
obs = .N
),
by = "Date"
]
dt2[,HouseClass := "c"]
dt2[obs < 300000,HouseClass := "b"]
dt2[obs < 200000,HouseClass := "a"]