def test_crosstab_mismatch_zones_values_shape():
zones = xa.DataArray(np.array([[1, 2]]))
values = xa.DataArray(np.array([[[1, 2, np.nan]]]),
dims=['lat', 'lon', 'race'])
values['race'] = ['cat1', 'cat2', 'cat3']
with pytest.raises(Exception) as e_info: # noqa
crosstab(zones_agg=zones, values_agg=values)
def test_crosstab_invalid_layer():
zones = xa.DataArray(np.array([[1, 2]]))
values = xa.DataArray(np.array([[[1, 2, np.nan]]]),
dims=['lat', 'lon', 'race'])
values['race'] = ['cat1', 'cat2', 'cat3']
layer = 'cat'
with pytest.raises(Exception) as e_info: # noqa
crosstab(zones_agg=zones, values_agg=values, layer=layer)
def test_crosstab_invalid_values():
zones = xa.DataArray(np.array([[1, 2, 0]], dtype=np.int))
# must be either int or float
values = xa.DataArray(np.array([[['apples', 'foobar', 'cowboy']]]),
dims=['lat', 'lon', 'race'])
values['race'] = ['cat1', 'cat2', 'cat3']
with pytest.raises(Exception) as e_info: # noqa
crosstab(zones_agg=zones, values_agg=values)
def test_crosstab_3d():
# expected results
crosstab_3d_results = {
'zone': [1, 2, 3],
'cat1': [6, 5, 6],
'cat2': [6, 5, 6],
'cat3': [6, 5, 6],
'cat4': [6, 5, 6],
}
# numpy case
zones_np, _, values_np = create_zones_values(backend='numpy')
df_np = crosstab(zones=zones_np,
values=values_np,
zone_ids=[1, 2, 3],
layer=-1)
# dask case
zones_da, _, values_da = create_zones_values(backend='dask')
df_da = crosstab(zones=zones_da,
values=values_da,
zone_ids=[1, 2, 3],
cat_ids=['cat1', 'cat2', 'cat3', 'cat4'],
layer=-1)
check_results(df_np, df_da, crosstab_3d_results)
# ----- no values case ------
crosstab_3d_novalues_results = {
'zone': [1, 2, 3],
'cat1': [0, 0, 0],
'cat2': [0, 0, 0],
'cat3': [0, 0, 0],
'cat4': [0, 0, 0],
}
# numpy case
zones_np, _, values_np = create_zones_values(backend='numpy')
df_np = crosstab(zones=zones_np,
values=values_np,
layer=-1,
zone_ids=[1, 2, 3],
nodata_values=1)
# dask case
zones_da, _, values_da = create_zones_values(backend='dask')
df_da = crosstab(zones=zones_da,
values=values_da,
layer=-1,
zone_ids=[1, 2, 3],
nodata_values=1)
check_results(df_np, df_da, crosstab_3d_novalues_results)
def test_crosstab_invalid_zones():
# invalid dims (must be 2d)
zones = xa.DataArray(np.array([1, 2, 0]))
values = xa.DataArray(np.array([[[1, 2, 0.5]]]),
dims=['lat', 'lon', 'race'])
values['race'] = ['cat1', 'cat2', 'cat3']
with pytest.raises(Exception) as e_info:
crosstab(zones_agg=zones, values_agg=values)
# invalid values (must be int)
zones = xa.DataArray(np.array([[1, 2, 0.5]]))
with pytest.raises(Exception) as e_info: # noqa
crosstab(zones_agg=zones, values_agg=values)
def test_crosstab_2d():
# count agg, expected results
crosstab_2d_results = {
'zone': [1, 2, 3],
0: [0, 0, 1],
1: [6, 0, 0],
2: [0, 4, 0],
}
# numpy case
zones_np, values_np, _ = create_zones_values(backend='numpy')
df_np = crosstab(
zones=zones_np,
values=values_np,
zone_ids=[1, 2, 3],
cat_ids=[0, 1, 2],
)
# dask case
zones_da, values_da, _ = create_zones_values(backend='dask')
df_da = crosstab(zones=zones_da,
values=values_da,
zone_ids=[1, 2, 3],
nodata_values=3)
check_results(df_np, df_da, crosstab_2d_results)
# percentage agg, expected results
crosstab_2d_percentage_results = {
'zone': [1, 2],
1: [100, 0],
2: [0, 100],
}
# numpy case
df_np = crosstab(zones=zones_np,
values=values_np,
zone_ids=[1, 2],
cat_ids=[1, 2],
nodata_values=3,
agg='percentage')
# dask case
df_da = crosstab(zones=zones_da,
values=values_da,
zone_ids=[1, 2],
cat_ids=[1, 2],
nodata_values=3,
agg='percentage')
check_results(df_np, df_da, crosstab_2d_percentage_results)
def test_crosstab_3d():
# create valid `values_agg` of np.nan and np.inf
values_agg = xa.DataArray(np.ones(24).reshape(2, 3, 4),
dims=['lat', 'lon', 'race'])
values_agg['race'] = ['cat1', 'cat2', 'cat3', 'cat4']
layer = 'race'
# create a valid `zones_agg` with compatiable shape
zones_arr = np.arange(6, dtype=np.int).reshape(2, 3)
zones_agg = xa.DataArray(zones_arr)
df = crosstab(zones_agg, values_agg, layer)
num_cats = len(values_agg.dims[-1])
# number of columns = number of categories
assert len(df.columns) == num_cats
# exclude region with 0 zone id
zone_idx = list(set(np.unique(zones_arr)) - {0})
num_zones = len(zone_idx)
# number of rows = number of zones
assert len(df.index) == num_zones
num_nans = df.isnull().sum().sum()
# no NaN
assert num_nans == 0
# values_agg are all 1s, so all categories have same percentage over zones
for col in df.columns:
assert len(df[col].unique()) == 1
df['check_sum'] = df.apply(
lambda r: r['cat1'] + r['cat2'] + r['cat3'] + r['cat4'], axis=1)
# sum of a row is 1.0
assert df['check_sum'][zone_idx[0]] == 1.0
def test_crosstab_no_values():
# create valid `values_agg` of 0s
values_agg = xa.DataArray(np.zeros(24).reshape(2, 3, 4),
dims=['lat', 'lon', 'race'])
values_agg['race'] = ['cat1', 'cat2', 'cat3', 'cat4']
# create a valid `zones_agg` with compatiable shape
zones_arr = np.arange(6, dtype=np.int).reshape(2, 3)
zones_agg = xa.DataArray(zones_arr)
df = crosstab(zones_agg, values_agg)
num_cats = len(values_agg.dims[-1])
# number of columns = number of categories
assert len(df.columns) == num_cats
# exclude region with 0 zone id
zone_idx = set(np.unique(zones_arr)) - {0}
num_zones = len(zone_idx)
# number of rows = number of zones
assert len(df.index) == num_zones
num_zeros = (df == 0).sum().sum()
# all are 0s
assert num_zeros == num_zones * num_cats
def test_crosstab_3d(backend, data_zones, data_values_3d, result_crosstab_3d):
layer, zone_ids, expected_result = result_crosstab_3d
df_result = crosstab(zones=data_zones,
values=data_values_3d,
zone_ids=zone_ids,
layer=layer)
check_results(backend, df_result, expected_result)
def test_crosstab_2d():
values_val = np.asarray([[0, 0, 10, 20],
[0, 0, 0, 10],
[np.inf, 30, 20, 50],
[10, 30, 40, 40],
[10, np.nan, 50, 0]])
values_agg = xa.DataArray(values_val, dims=['lat', 'lon'])
zones_val = np.asarray([[1, 1, 6, 6],
[1, 1, 6, 6],
[3, 5, 6, 6],
[3, 5, 7, 7],
[3, 7, 7, 0]])
zones_agg = xa.DataArray(zones_val, dims=['lat', 'lon'])
df = crosstab(zones_agg, values_agg)
num_cats = 6 # 0, 10, 20, 30, 40, 50
# number of columns = number of categories
assert len(df.columns) == num_cats
# exclude region with 0 zone id
zone_idx = list(set(np.unique(zones_agg.data)) - {0})
num_zones = len(zone_idx)
# number of rows = number of zones
assert len(df.index) == num_zones
df.loc[:, 'check_sum'] = df.sum(axis=1)
# sum of a row is 1.0
assert df['check_sum'][zone_idx[0]] == 1.0
def test_percentage_crosstab_2d(backend, data_zones, data_values_2d,
result_percentage_crosstab_2d):
nodata_values, zone_ids, cat_ids, expected_result = result_percentage_crosstab_2d
df_result = crosstab(zones=data_zones,
values=data_values_2d,
zone_ids=zone_ids,
cat_ids=cat_ids,
nodata_values=nodata_values,
agg='percentage')
check_results(backend, df_result, expected_result)
def test_count_crosstab_2d(backend, data_zones, data_values_2d,
result_count_crosstab_2d):
zone_ids, cat_ids, expected_result = result_count_crosstab_2d
df_result = crosstab(
zones=data_zones,
values=data_values_2d,
zone_ids=zone_ids,
cat_ids=cat_ids,
)
check_results(backend, df_result, expected_result)
def test_crosstab_3d_agg_method(backend, data_zones, data_values_3d,
result_crosstab_3d):
layer, zone_ids, expected_result = result_crosstab_3d
agg_methods = ['min', 'max', 'mean', 'sum', 'std', 'var', 'count']
for agg in agg_methods:
df_result = crosstab(zones=data_zones,
values=data_values_3d,
zone_ids=zone_ids,
layer=layer,
agg=agg)
check_results(backend, df_result, expected_result[agg])
def test_crosstab_invalid_input():
# invalid zones dims (must be 2d)
zones = xa.DataArray(np.array([1, 2, 0]))
values = xa.DataArray(np.array([[[1, 2, 0.5]]]),
dims=['lat', 'lon', 'race'])
values['race'] = ['cat1', 'cat2', 'cat3']
with pytest.raises(Exception) as e_info:
crosstab(zones_agg=zones, values_agg=values)
# invalid zones dtype (must be int)
zones = xa.DataArray(np.array([[1, 2, 0.5]]))
with pytest.raises(Exception) as e_info: # noqa
crosstab(zones_agg=zones, values_agg=values)
# invalid values
zones = xa.DataArray(np.array([[1, 2, 0]], dtype=np.int))
# values must be either int or float
values = xa.DataArray(np.array([[['apples', 'foobar', 'cowboy']]]),
dims=['lat', 'lon', 'race'])
values['race'] = ['cat1', 'cat2', 'cat3']
with pytest.raises(Exception) as e_info: # noqa
crosstab(zones_agg=zones, values_agg=values)
# mismatch shape zones and values
zones = xa.DataArray(np.array([[1, 2]]))
values = xa.DataArray(np.array([[[1, 2, np.nan]]]),
dims=['lat', 'lon', 'race'])
values['race'] = ['cat1', 'cat2', 'cat3']
with pytest.raises(Exception) as e_info: # noqa
crosstab(zones_agg=zones, values_agg=values)
# invalid layer
zones = xa.DataArray(np.array([[1, 2]]))
values = xa.DataArray(np.array([[[1, 2, np.nan]]]),
dims=['lat', 'lon', 'race'])
values['race'] = ['cat1', 'cat2', 'cat3']
# this layer does not exist in values agg
layer = 'cat'
with pytest.raises(Exception) as e_info: # noqa
crosstab(zones_agg=zones, values_agg=values, layer=layer)
def test_crosstab_2d_dtypes():
values_val = np.asarray(
[[0, 0, 10, 20], [0, 0, 0, 10], [np.inf, 30, 20, 50], [10, 30, 40, 40],
[10, np.nan, 50, 0]],
dtype=np.float16)
values_agg = xa.DataArray(values_val, dims=['lat', 'lon'])
zones_val = np.asarray([[1, 1, 6, 6], [1, 1, 6, 6], [3, 5, 6, 6],
[3, 5, 7, 7], [3, 7, 7, 0]])
zones_agg = xa.DataArray(zones_val, dims=['lat', 'lon'])
df = crosstab(zones_agg, values_agg)
assert isinstance(df, pd.DataFrame)
def test_crosstab_no_zones():
# create valid `values_agg`
values_agg = xa.DataArray(np.zeros(24).reshape(2, 3, 4),
dims=['lat', 'lon', 'race'])
values_agg['race'] = ['cat1', 'cat2', 'cat3', 'cat4']
# create a valid `zones_agg` with compatiable shape
# no zone
zones_arr = np.zeros((2, 3), dtype=np.int)
zones_agg = xa.DataArray(zones_arr)
num_cats = len(values_agg.dims[-1])
df = crosstab(zones_agg, values_agg)
# number of columns = number of categories
assert len(df.columns) == num_cats
# no row as no zone
assert len(df.index) == 0
