CODEWITHSUNDEEP
pythonr
jony
jony

Reputation: 934

rolling mean with quantile in R language

roll = df.rolling(window=3, center=False).quantile(0.5)

I am passing [72,73,74,71,75,72,77,75,72,79,73,72] as a DF

I get min 74 max 77

this code i did in Python

I want to do this in R

there are many libs in R like caTools and many more

runquantile(data[,2],3,probs = 0.5)(i tried this)

but no one is working properly

please help me out

Edit.

Based on the OP comment, here are data and code. This works with me.

DF <- c(72, 73, 74, 71, 75, 72, 77, 75, 72, 79, 73, 72)
runquantile(DF, 3, probs = 0.5)

Upvotes: 0

Views: 646

Answers (1)

r2evans
r2evans

Reputation: 160447

With almost no exception, anything "rolling" to me suggests the zoo package:

zoo::rollapply(DF, k=3, FUN=median) 
#  [1] 73 73 74 72 75 75 75 75 73 73
### this returns length(DF)-2*floor(width/2),
### problem if putting into a data.frame .. if so, then pad it:
zoo::rollapply(DF, 3, FUN=median, fill=NA)
#  [1] NA 73 73 74 72 75 75 75 75 73 73 NA

(I'll pad all from here on out.)

Quantiles:

zoo::rollapply(DF, width=3, quantile, probs=0.5, fill=NA)
#     50% 50% 50% 50% 50% 50% 50% 50% 50% 50%     
#  NA  73  73  74  72  75  75  75  75  73  73  NA

Quantiles with multiple probs returns a matrix:

zoo::rollapply(DF, width=3, quantile, probs=c(0.5,0.6), fill=NA)
#       50%  60%
#  [1,]  NA   NA
#  [2,]  73 73.2
#  [3,]  73 73.2
#  [4,]  74 74.2
#  [5,]  72 72.6
#  [6,]  75 75.4
#  [7,]  75 75.4
#  [8,]  75 75.4
#  [9,]  75 75.8
# [10,]  73 74.2
# [11,]  73 74.2
# [12,]  NA   NA

Same thing, but we can write a function (anonymous or named) that does whatever-we-need:

zoo::rollapply(DF, width=3, FUN = function(x) quantile(x, probs=c(0.5,0.6)), fill=NA)

For clarity, your statement "I get min 74 max 77" can be misleading. I get the full data:

In [15]: DF = pd.DataFrame({'B': [72,73,74,71,75,72,77,75,72,79,73,72]})

In [16]: DF.rolling(window=3, center=False).quantile(0.5)
Out[16]: 
       B
0    NaN
1    NaN
2   73.0
3   73.0
4   74.0
5   72.0
6   75.0
7   75.0
8   75.0
9   75.0
10  73.0
11  73.0

Since center=False, perhaps the call to zoo::rollapply above should use align=:

zoo::rollapply(DF, width=3, quantile, probs=0.5, fill = NA, align="right")
#         50% 50% 50% 50% 50% 50% 50% 50% 50% 50% 
#  NA  NA  73  73  74  72  75  75  75  75  73  73

The 50% labels are an artifact of using quantile and can be safely ignored and/or removed, as in

zoo::rollapply(DF, width=3, quantile, probs=0.5, fill = NA, align="right", names=FALSE)
#  [1] NA NA 73 73 74 72 75 75 75 75 73 73

Upvotes: 6

Related Questions