Washing Machine Project

Tracking progress

1. Machine lifetime & intensity

Key variables:

• machine_commissioning

• datapoint_time

• customer_cycles_amount

What to analyse:

• Machine age at service

→ datapoint_time – commissioning date

• Usage intensity:

→ cycles per year / per month

Calculation of machine age

Here I created a new data frame just to test it out if the code works. At the end I mutated a column in the original joined data frame. The age column is in days, and month column is in month.

machine.age <- join %>% 
  select(datapoint_time,machine_commissioning, machine_id) %>% 
  group_by(machine_id)

machine.age$dt <- as.vector(machine.age$datapoint_time)
machine.age$mt <- as.vector(machine.age$machine_commissioning)

machine.age <- machine.age %>% 
  group_by(machine_id) %>% 
  mutate(age = dt - mt, na.rm = TRUE)

machine.age <- machine.age %>% 
  group_by(machine_id) %>% 
  mutate(month = round(age/30.417))

join$month <- machine.age$month

library(knitr)
kable(machine.age[1:5,], caption = "New data frame to calculate age in days and months")

New data frame to calculate age in days and months

datapoint_time	machine_commissioning	machine_id	dt	mt	age	na.rm	month
2020-01-02	2014-06-20	201406209290	18263	16241	2022	TRUE	66
2020-01-02	2007-09-14	200709147860	18263	13770	4493	TRUE	148
2020-01-03	2007-06-17	200706173882	18264	13681	4583	TRUE	151
2020-01-06	2012-01-28	201201287469	18267	15367	2900	TRUE	95
2020-01-07	2009-03-20	200903202799	18268	14323	3945	TRUE	130

Scatterplot of age and cycles per month

Interesting questions:

1. Do failures depend on age or usage?

2. Are machines failing early or only after heavy use?

join <- join %>% 
  mutate(cycle_per_month = customer_cycles_amount / month, na.rm = TRUE)

ggplot(join, aes(month, cycle_per_month), theme(classic)) + 
  geom_point(size = 1)+
  geom_smooth()+
  labs(
    title = "Age and Usage Frequency",
    x = "Age in months",
    y = "Cycles per month"
  )

2. Usage behaviour

Key variables:

• machine_program_60degrees

• machine_program_40degrees

• machine_program_eco

• customer_household_type

What to analyse:

• Usage patterns:

• eco-heavy users vs high-temp users

Link to: household type