Generic Population Vector IF Improvements

[Samweli]

This Pull Request addresses #2543

The following are the changes are made

1. Changed exposure type from exposure area to exposure population.
2. Added action list and notes for the IF.
3. Changed exposure layer mode from classified to continuous.
4. Showing area name in impact_report instead of area_id
5. Added new table for showing number of population in each hazard level
6. Using the right value for population field in the impact layer

[timlinux]

Maybe area_name_field would make more sense here?

[timlinux]

@Samweli thanks for this patch. I have some mostly minor comments in the PR review above. If we can address these we can merge!

[Samweli]

Thanks @timlinux, i will look into them :).

[ismailsunni]

Thanks @Samweli

@samnawi can you try this PR?

[samnawi]

I've tried in my PC, both of Win7 64bit and Ubuntu 14.04 64bit.
I got some error message when I run vector population VS vector flood.

@Samweli I've a question for the vector population keyword wizard, What is the meaning of "Attribute for the type of the area"? Is the geomerty type or population type or what?

Thank you.

[Samweli]

Hi @samnawi, the type attribute stands for the area type(ward or district or sub-ward) and from the exposure data you have used, is the population values of type range? The IF expects integer type population values, this can be one of the reason it is raising an exception.

[ismailsunni]

Ouch, write long comment, and it's gone. Accidentally click back button.

• Is it right that the unit is density? Because in https://github.com/Samweli/inasafe/blob/develop/safe/impact_functions/generic/classified_polygon_people/impact_function.py#L513, there is multiplication between area ratio and the population. So, I think it should be population count.
• Please don't use hardcoded attribute name like in https://github.com/Samweli/inasafe/blob/develop/safe/impact_functions/generic/classified_polygon_people/impact_function.py#L384. The best practice would be get the field attribute from hazard layer's keywords
• Please use the similar approach that I use for classifying the hazard level. See https://github.com/inasafe/inasafe/pull/2546/files#diff-a53a88821aeba07646367921f6a9e767R112. Currently, you did like https://github.com/Samweli/inasafe/blob/develop/safe/impact_functions/generic/classified_polygon_people/impact_function.py#L402. It will be a problem if the layer has other attribute value, (e.g. HIGH LEVEL FLOOD)
• Can you put the table with hazard level in top of the impact report, then followed by the total affected population, then the breakdown of each area.

[Charlotte-Morgan]

Hi @ismailsunni - as requested more info on point 4 above.
a) the output layer is called "populations affected by each hazard zone" . This is then in the map legend. Please change the name of the default impact layer to "people affected by each hazard zone"
NOTE: this should be consistent with all hazard zone on people impact functions

b) the results in the output layer are displayed by the attribute safe_ag which is coded to mean yes or no. I think the results would be more meaningful if the label was the number of people in the hazard zone. No might be 0 and yes should be greater than 0.

c) The map title should also be "affected people" - with capitalisation as per standards.

[Charlotte-Morgan]

re PR #2600 & point 5 above - results still look dodgy to me. We now have negative unaffected people ?? The total numbers also look incorrect to me.

The notes and assumptions also need to be updated to use the same standard wording and changes made by @timlinux - i'm pretty sure we removed the human fractions statement previously

[Samweli]

Hi @Charlotte-Morgan thanks for your comments, i'm looking at the problems you have highlighted. Will get back once i have fix them.

[Charlotte-Morgan]

hi @ismailsunni @Samweli - I see that some of the issues above are addressed but the results are still confusing for me. I do not understand what you are trying to present with the results. I'm assuming that you are reporting on affected people. The number of people affected in the no zone is still HUGE.

Suggestion: Would it be possible to style the top of the report in a similar way to the flood vector on raster population and just show the number of people in the flood affected area (yes hazard zone)

This solves the HUGE value at no :)
@Samweli - if this works, you may want to think about what you want to show on the map; maybe its the number of people in the flood affected (yes) area? not just where the flooded area is?

Another issue I just noticed: is this IF calculating the min needs for all affected people or only a certain percentage? ie in flood vector on raster population we use the 1% rule but we also state this in the impact report and notes.
It would be useful to use similar notes if they are relevant:

@timlinux did a lot of work tidying up reports and notes & it would be great to get that work to persist in all impact reports (please).
Just leave out the second dot if you are not using a percentage. If there is a percentage state what it is.

Sorry - I think this gives you more things to tweak, but its getting there and it runs, even on data with NULL values so you have made great progress. :) And this IF will be so useful next week when it floods in Jakarta.