Subset by increasing count over time in R - r

Question

I have a data frame that contains count of sales by seller, year and month, called sales_by_month:
library(tidyverse)
sales_by_month <- tribble(
~Seller, ~Year, ~Month, ~Sales,
"John Doe", 2018, 1, 82,
"John Doe", 2018, 2, 72,
"John Doe", 2018, 3, 42,
"Sally Jane", 2018, 1, 25,
"Sally Jane", 2018, 2, 77)
I need to subset this dataset by only those sellers where their sales are increasing over time, and I cannot figure out how to do it.
The resulting subset dataset should contain;
Seller Year Month Sales
Sally Jane 2018 1 25
Sally Jane 2018 2 77
Because Sally's sales are increasing, while John's sales are decreasing.
Any help would be very much appreciated!!

Answer 1

First, make a variable representing the difference in Sales(I name it dif). If dif < 0, it means decreasing values exist in somebody's sales.
library(dplyr)
df %>% arrange(Seller, Year, Month) %>%
group_by(Seller) %>%
mutate(dif = c(0, diff(Sales))) %>%
filter(all(dif >= 0)) %>%
select(-dif) # drop dif
# Seller Year Month Sales
# <fct> <int> <int> <int>
# 1 SallyJane 2018 1 25
# 2 SallyJane 2018 2 77
More concise:
df %>% group_by(Seller) %>%
arrange(Seller, Year, Month) %>%
filter(all(c(0, diff(Sales)) >= 0))

Answer 2

How to do this really depends on how you want to define increasing over time. One way to define increasing over time is if there is a month over month increase. My solution just looks at if the last month had an increase but this could be changed to look at it different ways:
We calculate the change every month. We filter for just the last month and if it was a positive change. Then we pull out unique Seller names.
We filter for the seller names we obtained in part 1.
The code below does this as well as a dataframe we can load straight into R
library(tidyverse)
sales_by_month <- tribble(
~Seller, ~Year, ~Month, ~Sales,
"John Doe", 2018, 1, 82,
"John Doe", 2018, 2, 72,
"John Doe", 2018, 3, 42,
"Sally Jane", 2018, 1, 25,
"Sally Jane", 2018, 2, 77)
increased_from_last_month <- sales_by_month %>%
group_by(Seller) %>%
arrange(Seller, Year, Month) %>%
mutate(change = Sales - lag(Sales, default = 0)) %>%
summarise_all(last) %>%
filter(change > 0) %>%
pull(Seller) %>%
unique()
sales_by_month %>%
filter(Seller %in% increased_from_last_month)