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

Test fcNaniqr #71

Merged
merged 4 commits into from
Dec 13, 2024
Merged

Test fcNaniqr #71

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
cfd6464
added tests
j-emberton Dec 10, 2024
620041d
Update tests/unit_tests/test_ustar_cp/test_fcNaniqr.py
j-emberton Dec 10, 2024
be310da
fixed erroneous tests
j-emberton Dec 12, 2024
1685448
Merge branch '69-fcnaniqr' of github.com:Cambridge-ICCS/ONEFlux into …
j-emberton Dec 12, 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
34 changes: 24 additions & 10 deletions oneflux_steps/ustar_cp_refactor_wip/fcNaniqr.m
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
function [IQR]=fcnaniqr(X)
function [IQR]=fcNaniqr(X)

%fcnaniqr computes the interquartile range, ignoring NaNs.
%
Expand Down Expand Up @@ -29,29 +29,43 @@

% find non-singleton dimensions of length d

d=size(X); d=setdiff(d,1); nd=length(d);
d=size(X);
d=setdiff(d,1);
nd=length(d);

switch nd;
case 1;
iYaN=find(~isnan(X)); nYaN=length(iYaN); IQR=NaN;
if nYaN<=3;
y=X(iYaN); yN=prctile(y,25); yX=prctile(y,75); IQR=yX-yN;
iYaN=find(~isnan(X));
nYaN=length(iYaN);
IQR=NaN;
if nYaN<=3;
y=X(iYaN);
yN=prctile(y,25);
yX=prctile(y,75);
IQR=yX-yN;
end;
case 2;
[nr,nc]=size(X); IQR=NaN*ones(1,nc);
[nr,nc]=size(X);
IQR=NaN*ones(1,nc);
for ic=1:nc;
iYaN=find(~isnan(X(:,ic))); nYaN=length(iYaN);
iYaN=find(~isnan(X(:,ic)));
nYaN=length(iYaN);
if nYaN>3;
y=X(iYaN,ic); yN=prctile(y,25); yX=prctile(y,75); IQR(ic)=yX-yN;
end;
end;
case 3;
[nr,nc,nq]=size(X); IQR=NaN*ones(nc,nq);
[nr,nc,nq]=size(X);
IQR=NaN*ones(nc,nq);
for iq=1:nq;
for ic=1:nc;
iYaN=find(~isnan(X(:,ic,iq))); nYaN=length(iYaN);
iYaN=find(~isnan(X(:,ic,iq)));
nYaN=length(iYaN);
if nYaN>3;
y=X(iYaN,ic,iq); yN=prctile(y,25); yX=prctile(y,75); IQR(ic,iq)=yX-yN;
y=X(iYaN,ic,iq);
yN=prctile(y,25);
yX=prctile(y,75);
IQR(ic,iq)=yX-yN;
end;
end;
end;
Expand Down
146 changes: 146 additions & 0 deletions tests/unit_tests/test_ustar_cp/test_fcNaniqr.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,146 @@
import matlab
import numpy as np
import pytest

# Parameterized Test Cases
@pytest.mark.parametrize(
"vector, expected",
[
# Case 1: Array is 3 and has no NaNs
(matlab.double([[1], [2], [3]]), 1.5), # IQR of [1, 2, 3]

# Case 2: Array >3, contains NaNs, but has >3 real numbers
(matlab.double([[1], [2], [3], [4], [np.nan], [6]]), np.nan),

# Case 3: Array >3, contains NaNs, but has 3 or fewer real numbers
# Note this result is erroneous!
(matlab.double([[np.nan], [2], [3] ,[np.nan]]), 1.0),

# Case 4: Array >3 and has no NaNs
(matlab.double([[1], [2], [3], [4], [5]]), np.nan),
],
)
def test_fcnaniqr_vector_cases(matlab_engine, vector, expected):
"""
Test MATLAB's fcNaniqr function with different 1D vectors, including edge cases.
"""
# Call MATLAB function
result = matlab_engine.fcNaniqr(vector)

# Check for None result
assert result is not None, "Expected non-None result from MATLAB function"

# Check if result is correct (accounting for NaNs)
if np.isnan(expected):
assert np.isnan(result), f"Expected NaN, but got {result}"
else:
assert np.isclose(result, expected), f"Expected {expected}, got {result}"

@pytest.mark.parametrize(
"matrix, expected",
[
# Case 1a: No NaNs, Fully Populated, column length = 3
(matlab.double([[1, 2, 3], [4, 5, 6], [7, 8, 9]]), [np.nan, np.nan, np.nan]),

# Case 1b: No NaNs, Fully Populated, column length = 4
(matlab.double([[1, 2, 3], [4, 5, 6], [7, 8, 9], [10, 11, 12]]), [6.0, 6.0, 6.0]),

# Case 2a: Columns with NaNs, column length = 3
(matlab.double([[1, np.nan, 3], [4, np.nan, 6], [7, 8, 9]]), [np.nan, np.nan, np.nan]),

# Case 2b: Columns with NaNs, column length >3 (after nans ignored)
(matlab.double([[1, np.nan, 3], [4, np.nan, 6], [7, 8, 9], [10,11,12]]), [6.0, np.nan, 6.0]),

# Case 3: Columns with < 4 Real Rows
#Note weird behaviour of IQR estimation
(matlab.double([[1, np.nan], [2, 3]]), [1.5, np.nan]),

# Case 4: All NaNs
(matlab.double([[np.nan, np.nan], [np.nan, np.nan]]), [np.nan, np.nan]),
],
)
def test_fcnaniqr_2D_cases(matlab_engine, matrix, expected):
"""
Test MATLAB's fcNaniqr function with 2D matrices, including edge cases.
"""
# Call MATLAB function
result = matlab_engine.fcNaniqr(matrix)

# Verify Result
assert result is not None, "Expected non-None result from MATLAB function"

# Normalize result to ensure consistent access
if isinstance(result, float): # Scalar result
result = [[result]]
elif isinstance(result, list) and isinstance(result[0], float): # 1D result from MATLAB
result = [result]

# Check each column's IQR
for idx, exp_val in enumerate(expected):
# Access the correct result value
try:
result_value = result[0][idx] if len(result[0]) > idx else np.nan
except (IndexError, TypeError):
result_value = np.nan

# Perform the assertion
if np.isnan(exp_val):
assert np.isnan(result_value), f"Expected NaN, got {result_value} for column {idx}"
else:
assert np.isclose(result_value, exp_val), f"Expected {exp_val}, got {result_value} for column {idx}"

@pytest.mark.parametrize(
"matrix, expected",
[
# Case 1a: No NaNs, Fully Populated, 2 rows (<3)
(matlab.double([[[1, 2], [4, 5]], [[7, 8], [10, 11]]]), [[np.nan, np.nan]]),

# Case 1b: No NaNs, Fully Populated, 4 rows (>3)
(matlab.double([[[1, 2], [4, 5]],[[7, 8], [10, 11]],[[13, 14], [16, 17]],[[19, 20], [22, 23]]]), [[12.0, 12.0]]),

# Case 1b: Single NaN, Fully Populated, 4 rows (>3)
(matlab.double([[[np.nan, 2], [4, 5]],[[7, 8], [10, 11]],[[13, 14], [16, 17]],[[19, 20], [22, 23]]]), [[np.nan, 12.0]]),

# Case 1c: Single Layer 3D
(matlab.double([[[1, 2, 3]], [[4, 5, 6]], [[7, 8, 9]]]), [[np.nan, np.nan, np.nan]]),

# Case 2a: Contains NaNs
(matlab.double([[[1, np.nan], [4, 5]], [[np.nan, 8], [7, np.nan]]]), [[np.nan, np.nan]]),

# Case 2b: Mixed NaNs with Real
(matlab.double([[[1, np.nan], [4, 5]], [[7, 8], [10, 11]]]), [[np.nan, np.nan]]),

# Case 3: All NaNs
(matlab.double([[[np.nan, np.nan], [np.nan, np.nan]], [[np.nan, np.nan], [np.nan, np.nan]]]), [[np.nan, np.nan]]),
],
)
def test_fcnaniqr_3D_cases(matlab_engine, matrix, expected):
"""
Test MATLAB's fcNaniqr function with 3D matrices, including edge cases.
"""
# Call MATLAB function
result = matlab_engine.fcNaniqr(matrix)

# Verify Result
assert result is not None, "Expected non-None result from MATLAB function"

# Normalize result to ensure consistent access
if isinstance(result, float): # Scalar result
result = [[result]]
elif isinstance(result, list) and isinstance(result[0], float): # 1D result from MATLAB
result = [result]

# Check each column's IQR
for row_idx, expected_row in enumerate(expected):
for col_idx, exp_val in enumerate(expected_row):
# Access the correct result value
try:
result_value = result[row_idx][col_idx] if len(result[row_idx]) > col_idx else np.nan
except (IndexError, TypeError):
result_value = np.nan

# Perform the assertion
if np.isnan(exp_val):
assert np.isnan(result_value), f"Expected NaN, got {result_value} for row {row_idx}, col {col_idx}"
else:
assert np.isclose(result_value, exp_val), f"Expected {exp_val}, got {result_value} for row {row_idx}, col {col_idx}"