CODEWITHSUNDEEP
rsignal-processing
rahul
rahul

Reputation: 591

Working of stats::filter function R

I am seriously at a loss in front of this, what does the stats filter command in R do how is the "Convolution" method implemented. I understand the case when filter is 1 but for any other value of filter things get confusing. From another question in stackoverflow (simple examples of filter function, recursive option specifically) I understood how "convolution" works for filter =1 eg:

f1<-1,f2<-1,f3<-1
x<-c(1:5)
filter(x,c(f1,f2))
3  5  7  9 NA
#which translates to
x[1]*f1+x[2]*f2
x[2]*f1+x[3]*f2
x[3]*f1+x[4]*f2
x[4]*f1+x[5]*f2
x[5]*f1+x[6]*f2 #x[6] is NA
#filter other than 1
#filter<-c(1,2)
filter(x,c(1,2))
4  7 10 13 NA
#and not the below ones
x[1]*f1+x[2]*f2=5
x[2]*f1+x[3]*f2=8
x[3]*f1+x[4]*f2=11

and so on, what exactly is happening here? this might be trivial and because of the lack of understanding Convolution method, but I am not able to figure it out

Upvotes: 5

Views: 4688

Answers (2)

Trevor King
Trevor King

Reputation: 1

Measuring the time difference between Bluetooth beacon pings

Example using the stats filter

Where AP is the Access point, and Location is where the test object was measured from

Location DateTime Access Point MAC RSSI
Lab@(15m) 2021-02-25 12:04:34 "1838Y-1282400000" "dc:0d:08:00:06:09" -76
Lab@(15m) 2021-02-25 12:04:34 "1838Y-1258500000" "dc:0d:08:00:06:09" -75
.Many more rows omitted.. ... ... ... ...
lab-loadingdock 2021-02-25 12:29:10 "1838Y-1282400000" "dc:0d:08:00:06:09" -73
lab-loadingdock 2021-02-25 12:29:13 "1838Y-1283400000" "dc:0d:08:00:06:09" -79
lab-loadingdock 2021-02-25 12:29:13 "1838Y-1258600000" "dc:0d:08:00:06:09" -86
lab-loadingdock 2021-02-25 12:29:13 "1838Y-1282400000" "dc:0d:08:00:06:09" -73
lab-loadingdock 2021-02-25 12:29:13 "1838Y-1258500000" "dc:0d:08:00:06:09" -64
lab-loadingdock 2021-02-25 12:29:15 "1838Y-1282400000" "dc:0d:08:00:06:09" -76
LabOffice 2021-02-25 13:07:06 "1838Y-1258500000" "dc:0d:08:00:06:09" -76
LabOffice 2021-02-25 13:07:06 "1838Y-1282400000" "dc:0d:08:00:06:09" -92 

BLE_Beacon_data %>%
 dplyr::filter(AP=="1838Y-1283400000",Location=="lab-loadingdock")%>%
 pull(DateTime)%>%
 stats::filter(c(1,-1))

Upvotes: -1

Matthew Lundberg
Matthew Lundberg

Reputation: 42659

The filter is applied in reverse time order. So the first element of the second example is:

x[1]*2 + x[2]*1 = 2 + 2 = 4.

etc.

The definition of a convolution includes reversing the order of one of the inputs.

Upvotes: 4

Related Questions