Melb: check other PA mmet hours/week against survey

[carlhiggs]

@BelenZapata85 advised Cambridge team have been recalibrating approach to otherPA for Manchester, although it is not clear if this is reqd for Melbourne -- we should check before emulating changes.

As a preliminary check, otherPA mmet hours/wk summary statistics were generated by gender and age group for Melbourne using previously processed baseline results from earlier in the year using queries of Athena database tables,

Click to view sql

SELECT 
'< 20' age,
gender,
AVG(mmethr_othersport) as avg,
APPROX_PERCENTILE(mmethr_othersport, 0) as min, 
APPROX_PERCENTILE(mmethr_othersport, 0.25) p25, 
APPROX_PERCENTILE(mmethr_othersport, 0.50) p50, 
APPROX_PERCENTILE(mmethr_othersport, 0.75) p75, 
APPROX_PERCENTILE(mmethr_othersport, 1.0) max
FROM melbourne_base_pp_exposure_2018
WHERE age < 20
GROUP BY gender
union
SELECT 
'20 - 40' age,
gender,
AVG(mmethr_othersport) as avg,
APPROX_PERCENTILE(mmethr_othersport, 0) as min, 
APPROX_PERCENTILE(mmethr_othersport, 0.25) p25, 
APPROX_PERCENTILE(mmethr_othersport, 0.50) p50, 
APPROX_PERCENTILE(mmethr_othersport, 0.75) p75, 
APPROX_PERCENTILE(mmethr_othersport, 1.0) max
FROM melbourne_base_pp_exposure_2018
WHERE age >= 20 AND age < 40
GROUP BY gender
union
SELECT 
'40 - 60' age,
gender,
AVG(mmethr_othersport) as avg,
APPROX_PERCENTILE(mmethr_othersport, 0) as min, 
APPROX_PERCENTILE(mmethr_othersport, 0.25) p25, 
APPROX_PERCENTILE(mmethr_othersport, 0.50) p50, 
APPROX_PERCENTILE(mmethr_othersport, 0.75) p75, 
APPROX_PERCENTILE(mmethr_othersport, 1.0) max
FROM melbourne_base_pp_exposure_2018
WHERE age >= 40 AND age < 60
GROUP BY gender
union 
SELECT 
'60 - 80' age,
gender,
AVG(mmethr_othersport) as avg,
APPROX_PERCENTILE(mmethr_othersport, 0) as min, 
APPROX_PERCENTILE(mmethr_othersport, 0.25) p25, 
APPROX_PERCENTILE(mmethr_othersport, 0.50) p50, 
APPROX_PERCENTILE(mmethr_othersport, 0.75) p75, 
APPROX_PERCENTILE(mmethr_othersport, 1.0) max
FROM melbourne_base_pp_exposure_2018
WHERE age >= 60 AND age < 80
GROUP BY gender
union 
SELECT 
'80+' age,
gender,
AVG(mmethr_othersport) as avg,
APPROX_PERCENTILE(mmethr_othersport, 0) as min, 
APPROX_PERCENTILE(mmethr_othersport, 0.25) p25, 
APPROX_PERCENTILE(mmethr_othersport, 0.50) p50, 
APPROX_PERCENTILE(mmethr_othersport, 0.75) p75, 
APPROX_PERCENTILE(mmethr_othersport, 1.0) max
FROM melbourne_base_pp_exposure_2018
WHERE age >= 80
GROUP BY gender

The synthetic population distribution for other PA by age and gender looks roughly plausible at first glance - i.e. declines by age, with mmet hours/week trending slightly higher for males than females.

age	gender	avg	min	p25	p50	p75	max
< 20	female	7.4	0.0	0.0	11.8	13.5	14.8
< 20	male	9.5	0.0	0.0	14.8	17.3	18.6
20 - 40	female	6.8	0.0	0.0	8.0	13.4	14.8
20 - 40	male	9.0	0.0	0.0	11.8	17.4	18.6
40 - 60	female	5.4	0.0	0.0	3.0	11.9	12.0
40 - 60	male	7.4	0.0	0.0	4.8	15.7	15.8
60 - 80	female	4.3	0.0	0.0	0.3	10.1	11.9
60 - 80	male	6.1	0.0	0.0	0.5	13.9	15.7
80+	female	2.1	0.0	0.0	0.0	6.5	8.3
80+	male	3.3	0.0	0.0	0.0	10.4	12.1

I am copying these statistics here for reference, to check them against survey data when we get a moment.