Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Add spatial visualisations for decision matrices #893

Open
wants to merge 14 commits into
base: main
Choose a base branch
from
+293 −28
Open

Add spatial visualisations for decision matrices #893

Show file tree
Hide file tree
Changes from 8 commits
Commits
Show all changes
14 commits
Select commit Hold shift + click to select a range
6ef0899
Move `viz.decision_matrices` into `spatial.jl` and make it another in…
Rosejoycrocker Oct 25, 2024
2d9f37c
Fix how scores are plotted (should be inside loop), add use of human …
Rosejoycrocker Jan 22, 2024
25d2890
Update criteria map plotting method in docs + update example fig
Rosejoycrocker Oct 25, 2024
833a3c5
Updates and fix input types to spatial_map
Rosejoycrocker Oct 25, 2024
9d27f72
Split `rank_by_index` into another function `get_criteria_aggregate'
Rosejoycrocker Oct 25, 2024
05def0f
Updates to viz.map for plotting decision matrices for latest viz version
Rosejoycrocker Oct 25, 2024
2feceff
Update example to reflect latest workflow for plotting decision matrices
Rosejoycrocker Oct 25, 2024
7e26a3e
Formatting
Rosejoycrocker Oct 29, 2024
1c52cd3
Split into 2 functions
Rosejoycrocker Nov 4, 2024
2d18fa4
Change name of `get_criteria_aggregate` to `criteria_aggregate_scores`
Rosejoycrocker Nov 4, 2024
d2f4772
Update example and add new plotting functions to viz
Rosejoycrocker Nov 4, 2024
409bfba
Fix formatting
Rosejoycrocker Nov 4, 2024
01a997c
Add conversion to human readable text when plotting
Rosejoycrocker Nov 6, 2024
1a677cc
Add testing criteria visualisations to site selection tests
Rosejoycrocker Nov 13, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
44 changes: 22 additions & 22 deletions Manifest-v1.11.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
# This file is machine-generated - editing it directly is not advised

julia_version = "1.11.1"
julia_version = "1.11.0"
manifest_format = "2.0"
project_hash = "f5f80681e1f41a61ae2064c46066ad6e608da783"

Expand Down Expand Up @@ -71,9 +71,9 @@ version = "0.1.38"

[[deps.Adapt]]
deps = ["LinearAlgebra", "Requires"]
git-tree-sha1 = "6a55b747d1812e699320963ffde36f1ebdda4099"
git-tree-sha1 = "d80af0733c99ea80575f612813fa6aa71022d33a"
uuid = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"
version = "4.0.4"
version = "4.1.0"
weakdeps = ["StaticArrays"]

[deps.Adapt.extensions]
Expand Down Expand Up @@ -512,15 +512,15 @@ version = "0.1.10"

[[deps.DiskArrays]]
deps = ["LRUCache", "OffsetArrays"]
git-tree-sha1 = "dea23e3d787c25fbf5f879c4501da828bc729472"
git-tree-sha1 = "e0e89a60637a62d13aa2107f0acd169b9b9b77e7"
uuid = "3c3547ce-8d99-4f5e-a174-61eb10b00ae3"
version = "0.4.5"
version = "0.4.6"

[[deps.Distances]]
deps = ["LinearAlgebra", "Statistics", "StatsAPI"]
git-tree-sha1 = "66c4c81f259586e8f002eacebc177e1fb06363b0"
git-tree-sha1 = "c7e3a542b999843086e2f29dac96a618c105be1d"
uuid = "b4f34e82-e78d-54a5-968a-f98e89d6e8f7"
version = "0.10.11"
version = "0.10.12"
weakdeps = ["ChainRulesCore", "SparseArrays"]

[deps.Distances.extensions]
Expand Down Expand Up @@ -1104,10 +1104,10 @@ uuid = "88015f11-f218-50d7-93a8-a6af411a945d"
version = "4.0.0+0"

[[deps.LLVM]]
deps = ["CEnum", "LLVMExtra_jll", "Libdl", "Preferences", "Printf", "Requires", "Unicode"]
git-tree-sha1 = "4ad43cb0a4bb5e5b1506e1d1f48646d7e0c80363"
deps = ["CEnum", "LLVMExtra_jll", "Libdl", "Preferences", "Printf", "Unicode"]
git-tree-sha1 = "d422dfd9707bec6617335dc2ea3c5172a87d5908"
uuid = "929cbde3-209d-540e-8aea-75f648917ca0"
version = "9.1.2"
version = "9.1.3"

[deps.LLVM.extensions]
BFloat16sExt = "BFloat16s"
Expand Down Expand Up @@ -1218,9 +1218,9 @@ version = "1.6.0+0"

[[deps.Libgpg_error_jll]]
deps = ["Artifacts", "JLLWrappers", "Libdl"]
git-tree-sha1 = "fbb1f2bef882392312feb1ede3615ddc1e9b99ed"
git-tree-sha1 = "c6ce1e19f3aec9b59186bdf06cdf3c4fc5f5f3e6"
uuid = "7add5ba3-2f88-524e-9cd5-f83b8a55f7b8"
version = "1.49.0+0"
version = "1.50.0+0"

[[deps.Libiconv_jll]]
deps = ["Artifacts", "JLLWrappers", "Libdl"]
Expand Down Expand Up @@ -1285,9 +1285,9 @@ version = "1.11.0"

[[deps.LoggingExtras]]
deps = ["Dates", "Logging"]
git-tree-sha1 = "c1dd6d7978c12545b4179fb6153b9250c96b0075"
git-tree-sha1 = "f02b56007b064fbfddb4c9cd60161b6dd0f40df3"
uuid = "e6f89c97-d47a-5376-807f-9c37f3926c36"
version = "1.0.3"
version = "1.1.0"

[[deps.Lz4_jll]]
deps = ["Artifacts", "JLLWrappers", "Libdl"]
Expand Down Expand Up @@ -1954,9 +1954,9 @@ version = "1.2.1"

[[deps.SentinelArrays]]
deps = ["Dates", "Random"]
git-tree-sha1 = "ff11acffdb082493657550959d4feb4b6149e73a"
git-tree-sha1 = "305becf8af67eae1dbc912ee9097f00aeeabb8d5"
uuid = "91c51154-3ec4-41a3-a24f-3f23e20d615c"
version = "1.4.5"
version = "1.4.6"

[[deps.Serialization]]
uuid = "9e88b42a-f829-5b0c-bbe9-9e923198166b"
Expand Down Expand Up @@ -2075,9 +2075,9 @@ version = "0.1.1"

[[deps.StaticArrays]]
deps = ["LinearAlgebra", "PrecompileTools", "Random", "StaticArraysCore"]
git-tree-sha1 = "eeafab08ae20c62c44c8399ccb9354a04b80db50"
git-tree-sha1 = "777657803913ffc7e8cc20f0fd04b634f871af8f"
uuid = "90137ffa-7385-5640-81b9-e52037218182"
version = "1.9.7"
version = "1.9.8"
weakdeps = ["ChainRulesCore", "Statistics"]

[deps.StaticArrays.extensions]
Expand Down Expand Up @@ -2349,9 +2349,9 @@ version = "1.1.41+0"

[[deps.XZ_jll]]
deps = ["Artifacts", "JLLWrappers", "Libdl"]
git-tree-sha1 = "ac88fb95ae6447c8dda6a5503f3bafd496ae8632"
git-tree-sha1 = "15e637a697345f6743674f1322beefbc5dcd5cfc"
uuid = "ffd25f8a-64ca-5728-b0f7-c24cf3aae800"
version = "5.4.6+0"
version = "5.6.3+0"

[[deps.Xorg_libX11_jll]]
deps = ["Artifacts", "JLLWrappers", "Libdl", "Xorg_libxcb_jll", "Xorg_xtrans_jll"]
Expand Down Expand Up @@ -2421,9 +2421,9 @@ version = "0.7.3"

[[deps.YAXArrays]]
deps = ["CFTime", "DataStructures", "Dates", "DimensionalData", "DiskArrayTools", "DiskArrays", "Distributed", "DocStringExtensions", "Glob", "Interpolations", "IntervalSets", "IterTools", "Markdown", "OffsetArrays", "OnlineStats", "Optim", "OrderedCollections", "ParallelUtilities", "ProgressMeter", "Reexport", "Requires", "Statistics", "StatsBase", "Tables", "WeightedOnlineStats", "YAXArrayBase"]
git-tree-sha1 = "09cd01df4817c4d5f6eda845c4d75cc2855cf0b3"
git-tree-sha1 = "02d9149f67b26256ffa1e1c3915d474fae57099e"
uuid = "c21b50f5-aa40-41ea-b809-c0f5e47bfa5c"
version = "0.5.11"
version = "0.5.12"

[[deps.Zarr]]
deps = ["AWSS3", "Blosc", "CodecZlib", "DataStructures", "Dates", "DiskArrays", "HTTP", "JSON", "LRUCache", "OffsetArrays", "OpenSSL", "Pkg", "URIs", "ZipArchives"]
Expand Down
Binary file added docs/src/assets/imgs/criteria_spatial_plots.png
View file
Open in desktop
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
54 changes: 54 additions & 0 deletions docs/src/usage/analysis.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -188,6 +188,60 @@ save("ranks_plot.png", rank_fig)

![Rank frequency plots for multiple ranks](../assets/imgs/analysis/ranks_plot.png)

## Intervention location selection - plot criteria maps

```julia

mcda_funcs = ADRIA.decision.mcda_methods()

dom = ADRIA.load_domain("path to domain","45")

# Plot using weightings from first scenario
scens = ADRIA.sample_guided(dom, 2^2)
scen = scens[1, :]

# Get seeding preferences
seed_pref = ADRIA.decision.SeedPreferences(dom, scen)

# Calculate criteria vectors
# Cover
sum_cover = vec(sum(dom.init_coral_cover; dims=1).data)
# DHWS
dhw_scens = dom.dhw_scens[:, :, Int64(scen["dhw_scenario"])]
plan_horizon = Int64(scen["plan_horizon"])
decay = 0.99 .^ (1:(plan_horizon + 1)) .^ 2
dhw_projection = ADRIA.decision.weighted_projection(dhw_scens, 1, plan_horizon, decay, 75)
# Connectivity
area_weighted_conn = dom.conn.data .* ADRIA.site_k_area(dom)
conn_cache = similar(area_weighted_conn)
in_conn, out_conn, network = ADRIA.connectivity_strength(
area_weighted_conn, sum_cover, conn_cache
)

# Create decision matrix
seed_decision_mat = ADRIA.decision.decision_matrix(
dom.loc_ids,
seed_pref.names;
seed_in_connectivity=in_conn,
seed_out_connectivity=out_conn,
seed_heat_stress=dhw_projection,
seed_coral_cover=sum_cover
)

# Get results from applying MCDA algorithm
crit_agg = ADRIA.decision.get_criteria_aggregate(seed_pref, seed_decision_mat, mcda_funcs[1])

# Don't plot constant criteria
is_const = Bool[length(x) == 1 for x in unique.(eachcol(seed_decision_mat.data))]

# Plot normalized scores and criteria as map
fig = ADRIA.viz.map(dom, seed_decision_mat[criteria=.!is_const], crit_agg.scores./maximum(crit_agg.scores))
save("criteria_plots.png", fig)
```

![Spatial maps of location selection criteria](/ADRIA.jl/dev/assets/imgs/criteria_spatial_plots.png?raw=true "Spatial maps of location selection criteria")


### PAWN sensitivity (heatmap overview)

The PAWN sensitivity analysis method is a moment-independent approach to Global Sensitivity
Expand Down
78 changes: 76 additions & 2 deletions ext/AvizExt/viz/spatial.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -151,8 +151,9 @@ function create_map!(
end
end

# Remove any empty subplots
trim!(f)
if typeof(f) == GridLayout
trim!(f)
end

return f
end
Expand All @@ -167,6 +168,10 @@ Plot spatial choropleth of outcomes.
# Arguments
- `rs` : ResultSet
- `y` : results of scenario metric
- `S` : A normalised decision matrix calculated using decison.decision_matrices
- `scores` : Aggregated criteria scores.
- `criteria` : Names of criteria to be plotted, if not specified all criteria in
S will be plotted.
- `opts` : Aviz options
- `colorbar_label`, label for colorbar. Defaults to "Relative Cover"
- `color_map`, preferred colormap for plotting heatmaps
Expand Down Expand Up @@ -251,6 +256,75 @@ function ADRIA.viz.map!(
axis_opts
)
end
function ADRIA.viz.map(
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Since map is a very generic name, I don't think we should make a method called map plot a specific kind of map only (only maps of decision matrix scores). What I mean is: it is interesting to have a map function that plots multiple maps like you did, but not only for decision matrix. I could plot the average of some different metrics for each location, for example.

So, I propose we have an intermediate function selection_criteria_map, for example, that receives this matrix and this vector, concatenate them to become just a single matrix; builds up a tuple of titles for each plot; and call a map function that receives this matrix and the titles_tuple (and whatever else it needs) and plots this grid of maps. Then, in the future, if we want to plot other multiple map figure, it can be easily reused.

What do you think?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sounds good :)

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I prefer the current approach but I'm not 100% up on the context here, just latching onto the suggestion to create a separate map plotting function. Julia is not strictly an object oriented language and isn't intended to be used like Python or Java.

The advantage of having multiple dispatch (or something like it, as in R) is to prevent everyone from having to remember a lot of function names. Is it map or vizualize_selection_criteria_as_map or display_map or plot_selection_criteria_map or...

As a user I don't want to have to remember endless possible function names for specific use cases. I just want a map.

Instead, this behaviour of:

  1. You have a vector of results -> map() produces a single map.
  2. You have a matrix of results -> map() produces a series of maps.

is sensible to me.

But again, I don't have full context.

https://www.juliaopt.org/meetings/santiago2019/slides/stefan_karpinski.pdf

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I understand and I agree with what you said.

My problem with how this function is defined right now is that it doesn't plot selection criteria scores in particular. It plots multiple maps with whatever you pass as an argument. But still the names of variables and the plots default titles are related to this specific kind of plot (multiple maps with selection criteria scores). It's like having a function called sum and the arguments are called prime_number_a and prime_number_b. If it sums more than just prime numbers, they should be called number_a and number_b or whatever.

About the need for a specific function called selection_criteria_map, maybe we don't need one, I agree. depending on how this map function with multiple maps is defined we can just use this one and that's fine. My comment was more about making this map general than about the need of a specific function with a specific name for this kind of plot.

Copy link
Collaborator

@ConnectedSystems ConnectedSystems Nov 6, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Incidentally @Rosejoycrocker the variable names/titles displayed in the figure should be run through human_readable() (this isn't the exact function name but you should be able to find it easily)

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm not sure what the outcome of this discssuion is, but I've separated this into a generic map() method which plots multiple maps in a grid given a matrix of inputs, and then a function in location_selection.jl that plots criteria and an aggregate score given a decision matrix input and vector of scores (using the generic map() function). Not sure what you think but I think being able to quickly plot criteria and an output score is useful, hence its inclusion in location_selection.jl.

rs::Union{Domain,ResultSet},
M::YAXArray,
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

  1. This doesn't match the docstring.
  2. I am against naming variables with a single letter except in very specific cases. You have little to gain and everyone that reads you code (including your future self) have a lot of time to lose with this. :)

scores::Vector{Float64};
criteria::Vector{Symbol}=Array(M.criteria),
opts::OPT_TYPE=DEFAULT_OPT_TYPE(),
fig_opts::OPT_TYPE=set_figure_defaults(DEFAULT_OPT_TYPE()),
axis_opts::OPT_TYPE=set_axis_defaults(DEFAULT_OPT_TYPE())
)
f = Figure(; fig_opts...)
g = f[1, 1] = GridLayout()
ADRIA.viz.map!(
g, rs, M, scores; criteria=criteria, opts=opts, axis_opts=axis_opts
)
return f
end
function ADRIA.viz.map!(
g::Union{GridLayout,GridPosition},
rs::Union{Domain,ResultSet},
M::YAXArray,
scores::Vector{Float64};
criteria::Vector{Symbol}=Array(M.criteria),
opts::OPT_TYPE=DEFAULT_OPT_TYPE(),
axis_opts::OPT_TYPE=set_axis_defaults(DEFAULT_OPT_TYPE())
)
if length(rs.loc_data.site_id) != size(M, 1)
error("Only unfiltered decision matrices can be plotted.")
end

opts[:color_map] = get(opts, :color_map, :viridis)
opts[:colorbar_limits] = get(opts, :colorbar_limits, (0.0, 1.0))

m_spec = model_spec(rs)
criteria_names::Vector{String} = m_spec[
dropdims(
any(
reshape(criteria, 1, length(criteria)) .== m_spec[:, "fieldname"]; dims=2
);
dims=2
), "name"]
n_criteria::Int64 = length(criteria)
n_rows, n_cols = _calc_gridsize(n_criteria + 1)
step::Int64 = 1

for row in 1:n_rows, col in 1:n_cols
if step > length(criteria_names)
axis_opts[:title] = "Aggregate criteria score"
ADRIA.viz.map!(
g[row, col],
rs,
vec(scores);
opts=opts,
axis_opts=axis_opts
)
break
end
axis_opts[:title] = string(criteria_names[step])
ADRIA.viz.map!(
g[row, col],
rs,
vec(M[criteria=At(criteria[step])]);
opts=opts,
axis_opts=axis_opts
)

step += 1
end
return g
end

"""
ADRIA.viz.map(gdf::DataFrame; geom_col=:geometry, color=nothing)
Expand Down
28 changes: 25 additions & 3 deletions src/decision/Criteria/DecisionPreferences.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -131,6 +131,30 @@ Index of locations ordered by their rank
function rank_by_index(
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This function signature doesn't match the docstring (I know this wasn't updated by you, if you prefer just open an issue to fix that on a separate PR).

dp::T, dm::YAXArray, method::Union{Function,DataType}
)::Vector{Int64} where {T<:DecisionPreference}
res = get_criteria_aggregate(dp, dm, method)
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I imagine res comes from result, right? Everything any function returns is a result (the result of that function), so calling a variable result is a little bit vague (maybe with a few exceptions). Maybe calling it aggregate_score or something else that helps the reader would be better.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Its a result structure that the MCDA package uses, so it contains more than just the aggregate score. Maybe change it to decision_results or something?

is_maximal = res.bestIndex == argmax(res.scores)
return sortperm(res.scores; rev=is_maximal)
end

"""
get_criteria_aggregate(dp::T, dm::YAXArray, method::Function)::Tuple{Vector{Float64,Int64}}
where {T<:DecisionPreference}

Calculates raw aggreagted score for a set of locations, in order of the location indices.

Note: Ignores constant criteria values.

# Arguments
- `dp` : DecisionPreferences
- `dm` : The decision matrix to assess
- `method` : An MCDA method provided by the JMcDM package

# Returns
Returns raw aggreagted score for a set of locations
"""
function get_criteria_aggregate(
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

  1. You say this function returns a raw aggregated score for a set of locations, so maybe it could be called aggregated_scores or criteria_aggregated_scores? I just think this get part of the name may be a little redundant. And, since you are aggregating scores, adding the word scores at the end may clarify what this returns.

  2. The function signature doesn't match the docstring.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for spotting, I'll fix the signature and change the name to criteria_aggregated_scores

dp::T, dm::YAXArray, method::Union{Function,DataType}
) where {T<:DecisionPreference}
# Identify valid, non-constant, columns for use in MCDA
is_const = Bool[length(x) == 1 for x in unique.(eachcol(dm.data))]

Expand All @@ -150,9 +174,7 @@ function rank_by_index(
# or if the method fails for some reason...
throw(DomainError(res.scores, "No ranking possible"))
end

is_maximal = res.bestIndex == argmax(res.scores)
return sortperm(res.scores; rev=is_maximal)
return (scores=res.scores, bestIndex=res.bestIndex)
end

"""
Expand Down
Loading