Skip to content

Commit

Permalink
iss #381 merged dev
Browse files Browse the repository at this point in the history
  • Loading branch information
@BiancaMorandi
BiancaMorandi committed Mar 23, 2023
2 parents c718985 + 8b851ac commit bc3c510
Show file tree
Hide file tree
Showing 4 changed files with 170 additions and 81 deletions.
2 changes: 2 additions & 0 deletions CHANGELOG.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -37,6 +37,8 @@ _line_colors_, _legend_ and _figure_size_. (Issue #[349](https://github.com/brig
14. Addressed all Future and Deprecation warnings for matplotlib<=3.6.3, numpy<=1.24.1, pandas<=1.5.3. (Issue #[356](https://github.com/brightwind-dev/brightwind/issues/356)).
15. Update to work with versions 1.0 to 1.2 of IEA WIND Task 43 WRA Data Model (Issue #[306](https://github.com/brightwind-dev/brightwind/issues/306)).
16. Updated `LoadBrightHub` URL and generalised functions used for connecting to BrightHub platform without using `boto3` (Issue #[355](https://github.com/brightwind-dev/brightwind/issues/355)).
17. Removed hardcoded colours for `Shear.TimeOfDay` plots when `plot_type` is 'step' or 'line' and added a colour map. (Issue #[376](https://github.com/brightwind-dev/brightwind/issues/376)).
18: Fixed bug for `SpeedSort` where the `sector_predict` function was not interpolating data using two fit lines. (Issue #[377](https://github.com/brightwind-dev/brightwind/issues/377)).


## [2.0.0]
Expand Down
52 changes: 46 additions & 6 deletions brightwind/analyse/correlation.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -832,12 +832,18 @@ def __init__(self, ref_spd, target_spd, cutoff):
self.params = dict()
self.params['slope'] = (ymean2 - ymean1) / (xmean2 - xmean1)
self.params['offset'] = ymean1 - (xmean1 * self.params['slope'])
self.params['cutoff'] = cutoff
# print(self.params)

def sector_predict(self, x):
def linear_function(x, slope, offset):
return x * slope + offset
return x.transform(linear_function, slope=self.params['slope'], offset=self.params['offset'])
def bilinear_function(x, slope, offset, cutoff):
if x < cutoff:
return (slope * cutoff + offset) * x / cutoff # Line passing through zero
else:
return x * slope + offset # Line not passing through zero

return x.transform(bilinear_function, slope=self.params['slope'], offset=self.params['offset'],
cutoff=self.params['cutoff'])

def plot_model(self):
return plot_scatter(self.sector_ref,
Expand Down Expand Up @@ -1003,10 +1009,44 @@ def get_result_table(self):
result = result.sort_index()
return result

def plot(self):
def plot(self, figure_size=(10, 10.2)):
"""
Plots scatter plot of reference versus target speed data for each sector. The regression line and the line of
slope 1 passing through the origin are also shown on each plot.
:param figure_size: Figure size in tuple format (width, height)
:type figure_size: tuple
:returns: A matplotlib figure
:rtype: matplotlib.figure.Figure
**Example usage**
::
import brightwind as bw
data = bw.load_csv(bw.demo_datasets.demo_data)
m2_ne = bw.load_csv(bw.demo_datasets.demo_merra2_NE)
# Correlate by directional sector, using 36 sectors.
spd_srt_cor = bw.Correl.SpeedSort(m2_ne['WS50m_m/s'], m2_ne['WD50m_deg'],
data['Spd80mN'], data['Dir78mS'], averaging_prd='1D',
coverage_threshold=0.9, sectors=12)
spd_srt_cor.run()
# To plot the scatter subplots by directional sectors, the regression line and the line of
# slope 1 passing through the origin
spd_srt_cor.plot()
# To set the figure size
spd_srt_cor.plot(figure_size=(20, 20.2))
"""
predict = self.data[self._ref_spd_col_name] * np.nan
for model in self.speed_model:
self.speed_model[model].plot_model('Sector ' + str(model))
return self.plot_wind_directions()
predict = predict.combine_first(self.speed_model[model].sector_predict(self.speed_model[model].sector_ref))

return plot_scatter_by_sector(self.data[self._ref_spd_col_name], self.data[self._tar_spd_col_name],
self.data[self._ref_dir_col_name], trendline_y=predict,
sort_trendline_inputs=True, sectors=self.sectors,
line_of_slope_1=True, figure_size=figure_size)

@staticmethod
def _linear_interpolation(xa, xb, ya, yb, xc):
Expand Down
185 changes: 110 additions & 75 deletions brightwind/analyse/plot.py
View file
Original file line number Diff line number Diff line change
@@ -1,5 +1,4 @@
import matplotlib.pyplot as plt
from matplotlib.colors import LinearSegmentedColormap
import matplotlib as mpl
from matplotlib.dates import DateFormatter
import matplotlib.dates as mdates
Expand All @@ -16,7 +15,7 @@
import six
from colormap import rgb2hex, rgb2hls, hls2rgb
from matplotlib.ticker import PercentFormatter
from matplotlib.colors import ListedColormap, LinearSegmentedColormap
from matplotlib.colors import ListedColormap, to_hex, LinearSegmentedColormap
import warnings

register_matplotlib_converters()
Expand Down Expand Up @@ -56,10 +55,10 @@ def __init__(self):
self.secondary = '#2E3743' # asphalt, rgb[46/255, 55/255, 67/255]
self.tertiary = '#9B2B2C' # red'ish, rgb(155, 43, 44)
self.fourth = '#E57925' # orange'ish, rgb(229, 121, 37)
self.fifth = '#F2D869' # yellow'ish, rgb(242, 216, 105)
self.fifth = '#ffc008' # yellow'ish, rgb(255, 192, 8)
self.sixth = '#AB8D60'
self.seventh = '#A4D29F'
self.eighth = '#6E807B'
self.eighth = '#01958a'
self.ninth = '#3D636F' # blue grey
self.tenth = '#A49E9D'
self.eleventh = '#DA9BA6'
Expand All @@ -73,7 +72,16 @@ def __init__(self):
_col_map_colors = [self.primary_95, # lightest primary
self.primary, # primary
self.primary_10] # darkest primary
self._color_map = self._set_col_map(_col_map_colors)

_color_map_cyclical_colors = [self.secondary,
self.fifth,
self.primary,
self.tertiary,
self.secondary]

self._color_map = self._set_col_map('color_map', _col_map_colors)

self._color_map_cyclical = self._set_col_map('color_map_cyclical', _color_map_cyclical_colors)

self.color_list = [self.primary, self.secondary, self.tertiary, self.fourth, self.fifth, self.sixth,
self.seventh, self.eighth, self.ninth, self.tenth, self.eleventh, self.primary_35]
Expand All @@ -82,44 +90,52 @@ def __init__(self):
# mpl.rcParams['axes.prop_cycle']

@staticmethod
def _set_col_map(col_map_colors):
return LinearSegmentedColormap.from_list('color_map', col_map_colors, N=256)
def _set_col_map(color_map_name, col_map_colors):
return LinearSegmentedColormap.from_list(color_map_name, col_map_colors, N=256)

@property
def color_map(self):
return self._color_map

@property
def color_map_cyclical(self):
return self._color_map_cyclical

@color_map.setter
def color_map(self, col_map_colors):
self._color_map = self._set_col_map(col_map_colors)
self._color_map = self._set_col_map('color_map', col_map_colors)

@color_map_cyclical.setter
def color_map_cyclical(self, col_map_colors):
self._color_map_cyclical = self._set_col_map('color_map_cyclical', col_map_colors)


COLOR_PALETTE = _ColorPalette()


def _adjust_color_lightness(input_color, factor):
def _colormap_to_colorscale(cmap, n_colors):
"""
Function that transforms a matplotlib colormap to a list of colors
"""
return [to_hex(cmap(k*1/(n_colors-1))) for k in range(n_colors)]


def _adjust_color_lightness(r, g, b, factor):
"""
Generate the color corresponding to the input primary color corrected by a lightness or darkness defined by the
input factor percentage. Lighter colors are obtained with a factor >1 and darker colors with a factor <1.
:param input_color: Input primary color for deriving the adjusted color from. This accepts a string that represents
the name of the color (same input formats as matplotlib.colors.to_rgb()).
It can be an RGB or RGBA sequence or a string in any of several forms:
1) hex color string, like ‘#000FFF’
2) standard name, like ‘green’
3) letter from the set ‘rgbcmykw’
4) string representation of a float, like ‘0.4’, indicating gray on a 0-1 scale
:type input_color: RGB or RGBA sequence or str
:param factor: Factor defining the percentage of lightness (between 1 and 2) or darkness (between 0 and 1).
:type factor: float
:return: color in hex format
:rtype: hex
:param r: Intensity of red color between 0 and 255.
:type r: float
:param g: Intensity of green color between 0 and 255.
:type g: float
:param b: Intensity of blue color between 0 and 255.
:type b: float
:param factor: Factor defining the percentage of lightness (>1) or darkness (<1).
:type factor: float
:return: color in hex format
:rtype: hex
"""
if factor > 2 or factor < 0:
raise ValueError("The `factor` input must be a value between 0 and 2.")

r, g, b = tuple(255 * np.array(mpl.colors.to_rgb(input_color)))

hue, lightness, saturation = rgb2hls(r / 255.0, g / 255.0, b / 255.0)
lightness = max(min(lightness * factor, 1.0), 0.0)
r, g, b = hls2rgb(hue, lightness, saturation)
Expand Down Expand Up @@ -822,38 +838,42 @@ def plot_scatter_wspd(x_wspd_series, y_wspd_series, x_label=None, y_label=None,
return scat_plot


def plot_scatter_by_sector(x, y, wdir, trendline_y=None, line_of_slope_1=True, sectors=12,
def plot_scatter_by_sector(x, y, wdir, trendline_y=None, sort_trendline_inputs=False, line_of_slope_1=True, sectors=12,
x_limits=None, y_limits=None, axes_equal=True, figure_size=(10, 10.2), **kwargs):
"""
Plot scatter subplots (with shared x and y axis) of x versus y for each directional sector. If a trendline
timeseries is given as input then this is also plotted in the graph. The line with slope 1 and passing
through the origin is shown if line_of_slope_1=True
:param x: The x-axis values or reference variable.
:type x: pd.Series
:param y: The y-axis values or target variable.
:type y: pd.Series
:param wdir: Timeseries of wind directions.
:type wdir: pd.Series
:param trendline_y: Series of trendline y values.
:type trendline_y: pd.Series
:param line_of_slope_1: Boolean to choose to plot the line with slope one and passing through the origin.
:type line_of_slope_1: Bool
:param sectors: Number of directional sectors
:type sectors: int
:param x_limits: x-axis min and max limits. Can be set to None to let the code derive the min and max from
the x_wspd_series.
:type x_limits: tuple, None
:param y_limits: y-axis min and max limits. Can be set to None to let the code derive the min and max from
the y_wspd_series.
:type y_limits: tuple, None
:param axes_equal: Boolean to set the units for the x and y axes to be equal. If x_limits and y_limits are
both None then the two axes limits are set to be the same.
:type axes_equal: Bool
:param figure_size: Figure size in tuple format (width, height)
:type figure_size: tuple
:param kwargs: Additional keyword arguments for matplotlib.pyplot.subplot
:returns: matplotlib.figure.Figure
:param x: The x-axis values or reference variable.
:type x: pd.Series
:param y: The y-axis values or target variable.
:type y: pd.Series
:param wdir: Timeseries of wind directions.
:type wdir: pd.Series
:param trendline_y: Series of trendline y values. This needs to be derived using the x-axis timeseries
values as it is plotted against x.
:type trendline_y: pd.Series
:param sort_trendline_inputs: Boolean to chose if trendline inputs should be sorted in ascending order. Default is
False and trendline inputs are not sorted.
:type sort_trendline_inputs: Bool
:param line_of_slope_1: Boolean to choose to plot the line with slope one and passing through the origin.
:type line_of_slope_1: Bool
:param sectors: Number of directional sectors
:type sectors: int
:param x_limits: x-axis min and max limits. Can be set to None to let the code derive the min and max
from the x_wspd_series.
:type x_limits: tuple, None
:param y_limits: y-axis min and max limits. Can be set to None to let the code derive the min and max
from the y_wspd_series.
:type y_limits: tuple, None
:param axes_equal: Boolean to set the units for the x and y axes to be equal. If x_limits and y_limits
are both None then the two axes limits are set to be the same.
:type axes_equal: Bool
:param figure_size: Figure size in tuple format (width, height)
:type figure_size: tuple
:param kwargs: Additional keyword arguments for matplotlib.pyplot.subplot
:returns: matplotlib.figure.Figure
**Example usage**
::
Expand Down Expand Up @@ -920,7 +940,13 @@ def plot_scatter_by_sector(x, y, wdir, trendline_y=None, line_of_slope_1=True, s
else:
trendline_y_input = trendline_y

_scatter_subplot(x[logic_sect], y[logic_sect], trendline_y_input, trendline_x=None,
if sort_trendline_inputs:
trendline_x_input = sorted([x_data for x_data in x[logic_sect]])
trendline_y_input = sorted([y_data for y_data in trendline_y_input])
else:
trendline_x_input = x[logic_sect]

_scatter_subplot(x[logic_sect], y[logic_sect], trendline_y_input, trendline_x=trendline_x_input,
line_of_slope_1=line_of_slope_1, x_label=None, y_label=None,
x_limits=x_limits, y_limits=y_limits, axes_equal=axes_equal,
subplot_title=str(round(ratio_min)) + '-' + str(round(ratio_max)),
Expand Down Expand Up @@ -1191,11 +1217,12 @@ def _bar_subplot(data, x_label=None, y_label=None, min_bar_axis_limit=None, max_
bar_color = bar_colors[i]
# The offset in x direction of that bar
x_offset = (i - n_bars / 2) * bar_width + bar_width / 2
r, g, b = tuple(255 * np.array(mpl.colors.to_rgb(bar_color))) # hex to rgb format

for data_bar, data_bin in zip(data[name], data_bins):
if vertical_bars:
ax.imshow(np.array([[mpl.colors.to_rgb(bar_color)],
[mpl.colors.to_rgb(_adjust_color_lightness(bar_color, factor=1.8))]]),
[mpl.colors.to_rgb(_adjust_color_lightness(r, g, b, factor=1.8))]]),
interpolation='gaussian', extent=(data_bin + x_offset - bar_width / 2,
data_bin + x_offset + bar_width / 2, 0,
data_bar),
Expand All @@ -1204,7 +1231,7 @@ def _bar_subplot(data, x_label=None, y_label=None, min_bar_axis_limit=None, max_
edgecolor=bar_color, linewidth=line_width, fill=False,
zorder=1)#5
else:
cmp = _create_colormap(mpl.colors.to_rgb(_adjust_color_lightness(bar_color, factor=1.8)),
cmp = _create_colormap(mpl.colors.to_rgb(_adjust_color_lightness(r, g, b, factor=1.8)),
mpl.colors.to_rgb(bar_color))
ax.imshow(_gradient_image(direction=1, cmap_range=(0, 1)), cmap=cmp,
interpolation='gaussian',
Expand Down Expand Up @@ -1891,40 +1918,48 @@ def plot_log_law(avg_slope, avg_intercept, wspds, heights, max_plot_height=None)


def plot_shear_time_of_day(df, calc_method, plot_type='step'):
"""
Function used by Shear.TimeOfDay for plotting the hourly shear for each calendar month or an average of all months.
The color map used for plotting the shear by time of day for each calendar month depends on the plot_type input:
1) if 'line' 'step' the COLOR_PALETTE.color_map_cyclical is used
2) if '12x24' the 'COLOR_PALETTE.color_map is used
The color used for plotting the average of all months shear is COLOR_PALETTE.primary.
:param df: Series of average shear by time of day or DataFrame of shear by time of day for
each calendar month.
:type df: pandas.Series or pandas.DataFrame
:param calc_method: Method used by Shear.TimeOfDay for shear calculation, either 'power_law' or 'log_law'.
Input used for defining label of y axis.
:type calc_method: str
:param plot_type: Type of plot to be generated. Options include 'line', 'step' and '12x24'. Default is 'step'.
:type plot_type: str
:returns: A shear by time of day plot
"""
df_copy = df.copy()
# colours in use
colors = [(0.6313725490196078, 0.6470588235294118, 0.6705882352941176, 1.0), # Jan
(0.1568627450980392, .19215686274509805, 0.6705882352941176, 1.0), # Feb
(0.06666666666666667, 0.4196078431372549, 0.6901960784313725, 1.0), # March
(0.22745098039215686, 0.7294117647058823, 0.9803921568627451, 1.0), # April
(0.2392156862745098, 0.5666666666666667, 0.42745098039215684, 1.0), # May
(0.4117647058823529, 0.7137254901960784, 0.16470588235294117, 1.0), # June
(0.611764705882353, 0.7725490196078432, 0.21568627450980393, 1.0), # July
(0.6823529411764706, 0.403921568627451, 0.1607843137254902, 1.0), # Aug
(0.7901960784313726, 0.48627450980392156, 0.1843137254901961, 1.0), # Sep
(1, 0.7019607843, .4, 1), # Oct
(0, 0, 0, 1.0), # Nov
(0.40588235294117647, 0.43137254901960786, 0.4666666666666667, 1.0)] # Dec

if len(df.columns) == 1:
colors[0] = colors[5]

if calc_method == 'power_law':
label = 'Average Shear'

if calc_method == 'log_law':
elif calc_method == 'log_law':
label = 'Roughness Coefficient'
else:
label = calc_method

if plot_type == '12x24':
df.columns = np.arange(1, 13, 1)
df.index = np.arange(0, 24, 1)
df[df.columns[::-1]]
return plot_12x24_contours(df, label=(label, 'mean'), plot='tod')

else:
colors = _colormap_to_colorscale(COLOR_PALETTE.color_map_cyclical, 13)
colors = colors[:-1]
if len(df.columns) == 1:
colors[0] = COLOR_PALETTE.primary

fig, ax = plt.subplots(figsize=(10, 10))
ax.set_xlabel('Time of Day')
ax.set_ylabel(label)
import matplotlib.dates as mdates

# create x values for plot
idx = pd.date_range('2017-01-01 00:00', '2017-01-01 23:00', freq='1H').time
Expand Down

0 comments on commit bc3c510

Please sign in to comment.