POP247
Google Places
Radoslaw Panczak
24 June, 2022
1 Introduction
These are preliminary results from the scraped Google Places data for Brisbane area.
Place is a location uniquely defined by Google, and can include places in the same physical location (ie. duplicated coordinates).
The results focus on places where ‘popular times’ were available.
The exact algorithm on how this ‘popularity’ is determined is, of course, unknown.
Best case scenario? Caltex Woolworths Stafford petrol station
2 Raw data
Raw dataset consists of:
- 41165 records, with each record being a ‘place’ with unique Google ID
- 26772 unique locations (defined as lat long pairs)
- 11 levels of place description
- Exact time, relative time and content of reviews of the place
- ‘popular times’ values, coded on hourly basis for each day of the week, ie. (potentially) 168 values for each place
3 Derived variables
Variables derived from original values include:
popular_complete- binary indicator of completeness of popular times, where all 168 vlues are filledpopular_any- categorical indicator of popular times availability (values: None, Some, Complete)popular_filled- percentage of popularity timepoints with values (0%-100% range, 168 times filled = 100%)popular_meansmean popular time of a place (overall)- Original time range of 0-167 was recoded into days of the week, time
of the day and combination of the above (for instance:
Fri 20:00-20:59) - Dataset was also saved in ‘tidy’ format of one observation per place and time.
4 No ‘popular times’?
5 Availability of ‘popular times’
Only small amount of places has any information on popularity, with a tiny fraction having complete info:
popular_any <categorical>
# total N=41165 valid N=41165 mean=1.04 sd=0.22
Value | N | Raw % | Valid % | Cum. %
-------------------------------------------
None | 39438 | 95.80 | 95.80 | 95.80
Some | 1631 | 3.96 | 3.96 | 99.77
Complete | 96 | 0.23 | 0.23 | 100.00
<NA> | 0 | 0.00 | <NA> | <NA>
Note: From this point forward only places with
any information on popular times will be analyzed.
6 Big variety of types of places (1)
x <character>
# total N=1727 valid N=1727 mean=37.59 sd=21.24
Value | N | Raw % | Valid % | Cum. %
--------------------------------------------------------
restaurant | 337 | 19.51 | 19.51 | 19.51
cafe | 200 | 11.58 | 11.58 | 31.09
bar | 159 | 9.21 | 9.21 | 40.30
point_of_interest | 157 | 9.09 | 9.09 | 49.39
park | 79 | 4.57 | 4.57 | 53.97
gym | 72 | 4.17 | 4.17 | 58.14
grocery_or_supermarket | 54 | 3.13 | 3.13 | 61.26
health | 46 | 2.66 | 2.66 | 63.93
store | 45 | 2.61 | 2.61 | 66.53
food | 34 | 1.97 | 1.97 | 68.50
meal_takeaway | 33 | 1.91 | 1.91 | 70.41
gas_station | 32 | 1.85 | 1.85 | 72.26
shopping_mall | 32 | 1.85 | 1.85 | 74.12
car_dealer | 29 | 1.68 | 1.68 | 75.80
meal_delivery | 28 | 1.62 | 1.62 | 77.42
train_station | 27 | 1.56 | 1.56 | 78.98
transit_station | 22 | 1.27 | 1.27 | 80.25
electronics_store | 20 | 1.16 | 1.16 | 81.41
bakery | 17 | 0.98 | 0.98 | 82.40
department_store | 17 | 0.98 | 0.98 | 83.38
car_repair | 16 | 0.93 | 0.93 | 84.31
clothing_store | 16 | 0.93 | 0.93 | 85.23
liquor_store | 13 | 0.75 | 0.75 | 85.99
convenience_store | 12 | 0.69 | 0.69 | 86.68
home_goods_store | 12 | 0.69 | 0.69 | 87.38
furniture_store | 11 | 0.64 | 0.64 | 88.01
parking | 11 | 0.64 | 0.64 | 88.65
pharmacy | 11 | 0.64 | 0.64 | 89.29
bank | 10 | 0.58 | 0.58 | 89.87
bicycle_store | 10 | 0.58 | 0.58 | 90.45
library | 9 | 0.52 | 0.52 | 90.97
night_club | 9 | 0.52 | 0.52 | 91.49
doctor | 8 | 0.46 | 0.46 | 91.95
car_rental | 7 | 0.41 | 0.41 | 92.36
hardware_store | 7 | 0.41 | 0.41 | 92.76
movie_theater | 7 | 0.41 | 0.41 | 93.17
travel_agency | 7 | 0.41 | 0.41 | 93.57
book_store | 6 | 0.35 | 0.35 | 93.92
dentist | 6 | 0.35 | 0.35 | 94.27
hospital | 6 | 0.35 | 0.35 | 94.61
pet_store | 6 | 0.35 | 0.35 | 94.96
post_office | 6 | 0.35 | 0.35 | 95.31
shoe_store | 6 | 0.35 | 0.35 | 95.66
veterinary_care | 6 | 0.35 | 0.35 | 96.00
car_wash | 5 | 0.29 | 0.29 | 96.29
general_contractor | 5 | 0.29 | 0.29 | 96.58
hair_care | 5 | 0.29 | 0.29 | 96.87
museum | 5 | 0.29 | 0.29 | 97.16
storage | 5 | 0.29 | 0.29 | 97.45
art_gallery | 4 | 0.23 | 0.23 | 97.68
beauty_salon | 4 | 0.23 | 0.23 | 97.92
church | 4 | 0.23 | 0.23 | 98.15
physiotherapist | 4 | 0.23 | 0.23 | 98.38
finance | 3 | 0.17 | 0.17 | 98.55
laundry | 3 | 0.17 | 0.17 | 98.73
local_government_office | 3 | 0.17 | 0.17 | 98.90
accounting | 2 | 0.12 | 0.12 | 99.02
atm | 2 | 0.12 | 0.12 | 99.13
bowling_alley | 2 | 0.12 | 0.12 | 99.25
insurance_agency | 2 | 0.12 | 0.12 | 99.36
jewelry_store | 2 | 0.12 | 0.12 | 99.48
amusement_park | 1 | 0.06 | 0.06 | 99.54
lawyer | 1 | 0.06 | 0.06 | 99.59
moving_company | 1 | 0.06 | 0.06 | 99.65
natural_feature | 1 | 0.06 | 0.06 | 99.71
place_of_worship | 1 | 0.06 | 0.06 | 99.77
police | 1 | 0.06 | 0.06 | 99.83
real_estate_agency | 1 | 0.06 | 0.06 | 99.88
spa | 1 | 0.06 | 0.06 | 99.94
university | 1 | 0.06 | 0.06 | 100.00
<NA> | 0 | 0.00 | <NA> | <NA>7 Big variety of types of places (2)
There are 70 different labels of places, and that is just for the
types.0 variable. On top of that there are other levels of
type varables introducing man combinations.
Clearly that typology would have to simplified if some further use is
planned. Many places refer to similar entities, for instance
bicycle_store & pet_store could be both
treated as SHOP category, whereas bank and
dentist as SERVICES (?).
There are also issues of data quality, for instance ‘Hungry Jack’s’
is classfied sometimes as restaurant sometimes as
meal_takeaway.
Also the quality of point_of_interest and
food classes is rather poor - mixing a lot of diverse types
of places such as bars, parks, shops and services.
8 Quality issues
All the places with ‘hotel’ in the name, not that many are
types.0 ‘hotels’:
9 Simplifying typology?
[1] 27type <character>
# total N=1727 valid N=1712 mean=17.10 sd=6.54
Value | N | Raw % | Valid % | Cum. %
------------------------------------------------
restaurant | 329 | 19.05 | 19.22 | 19.22
SHOP | 308 | 17.83 | 17.99 | 37.21
SERVICES | 252 | 14.59 | 14.72 | 51.93
cafe | 206 | 11.93 | 12.03 | 63.96
bar | 164 | 9.50 | 9.58 | 73.54
SPORT | 110 | 6.37 | 6.43 | 79.96
park | 84 | 4.86 | 4.91 | 84.87
meal_takeaway | 73 | 4.23 | 4.26 | 89.14
gas_station | 32 | 1.85 | 1.87 | 91.00
train_station | 27 | 1.56 | 1.58 | 92.58
transit_station | 22 | 1.27 | 1.29 | 93.87
shopping_mall | 19 | 1.10 | 1.11 | 94.98
bakery | 17 | 0.98 | 0.99 | 95.97
liquor_store | 12 | 0.69 | 0.70 | 96.67
night_club | 9 | 0.52 | 0.53 | 97.20
AIRPORT | 8 | 0.46 | 0.47 | 97.66
CULTURE | 7 | 0.41 | 0.41 | 98.07
movie_theater | 7 | 0.41 | 0.41 | 98.48
parking | 6 | 0.35 | 0.35 | 98.83
church | 5 | 0.29 | 0.29 | 99.12
atm | 4 | 0.23 | 0.23 | 99.36
ENTERTAINMENT | 4 | 0.23 | 0.23 | 99.59
bowling_alley | 2 | 0.12 | 0.12 | 99.71
BRIDGE | 2 | 0.12 | 0.12 | 99.82
INT_CAFE | 2 | 0.12 | 0.12 | 99.94
amusement_park | 1 | 0.06 | 0.06 | 100.00
<NA> | 15 | 0.87 | <NA> | <NA>10 Completeness of ‘popular times’
Places with popular times cluster in the center of Brisbane:
Some Complete
Brisbane - East 0 1
Brisbane - North 156 10
Brisbane - South 101 3
Brisbane - West 86 7
Brisbane Inner City 1281 75
Moreton Bay - South 7 011 Missings in ‘popular times’
12 Filled values distribution
Note: dashed line is for mean (and its CI)
13 Filled values across type of place
Note: naive linear regression of completness across categorical place type.
14 Filled values across space (1)
