def __init__(
self,
x,
y,
z,
data=None,
fillna=None,
ax=None,
colorbar="vertical",
im_kwargs={},
):
self.ax = get_ax(ax)
if data is None:
data = pd.DataFrame({"x": x, "y": y, "z": z})
x, y, z = "x", "y", "z"
data = (data.reset_index().set_index(
[y, x]).unstack(x)[z].sort_index(ascending=False))
if fillna is not None:
data.fillna(fillna)
bounds = data.columns[0], data.columns[-1], data.index[-1], data.index[
0]
self.im = self.ax.imshow(data.values, extent=(bounds), **im_kwargs)
if colorbar:
if colorbar not in ["vertical", "horizontal"]:
colorbar = "vertical"
self.cbar = add_colorbar(self.im, orientation=colorbar)
else:
self.cbar = None
self.data = data
def plot(self, ax=None, logx=False, logy=False):
ax = get_ax(ax)
self.power.abs().plot(ax=ax)
ax.set_xlim([0, self.nyquist])
if logx:
ax.set_xscale("log")
if logy:
ax.set_yscale("log")
return ax
def plot_PSTH(data,
ax=None,
plot='line',
plot_params={},
PSTH_params={},
**kwargs):
ax = get_ax(ax)
xticks, counts = PSTH(**PSTH_params).compute(data,
nrows=kwargs.get('nrows'))
if plot == 'line':
plot_params = ChainMap(plot_params, default_PSTHplot_line_params)
ax.plot(xticks, counts, **plot_params)
return ax
def Regression(
x,
y,
data=None,
x_pred=None,
drop_na=True,
fitline=True,
scatter=True,
engine="matplotlib",
ax=None,
fitline_kwargs={},
scatter_kwargs={},
):
""" A convenience function for generating a regression plot
Parameters
----------
x, y, data, drop_na
see simi.analysis.linear_regression.LinearRegression
fitline: bool; default=True
determines if a fitline is plotted
scatter: bool; default=True
determines if the original data is plotted
engine: str; default='matplotlib'
ax: Axes; default=None
fitline_kwargs: dict-like; default={}
scatter_kwargs: dict-like; default={}
"""
regression_output = LinearRegression(x, y, data, drop_na, x_pred=x_pred)
if engine == "holoviews":
plot = _regression_holoviews(regression_output, fitline, scatter)
return plot
elif engine == "matplotlib":
ax = get_ax(ax)
if fitline:
ax.plot(regression_output.x_pred, regression_output.y_pred,
**fitline_kwargs)
if scatter:
ax.scatter(
regression_output.data[regression_output.x_label],
regression_output.data[regression_output.y_label],
**scatter_kwargs,
)
return ax
else:
raise ValueError(
f"Engine not implemented: {engine}. Choose one of: ['holoviews','matplotlib']"
)
def __init__(self, x, y, z, data=None, fillna=None, ax=None, colorbar='vertical', im_kwargs={}):
self.ax = get_ax(ax)
if data is None:
data = pd.DataFrame({'x':x, 'y':y, 'z':z})
x, y, z = 'x', 'y', 'z'
data = data.reset_index().set_index([y, x]).unstack(x)[z].sort_index(ascending=False)
if fillna is not None:
data.fillna(fillna)
bounds = data.columns[0], data.columns[-1], data.index[-1], data.index[0]
self.im = self.ax.imshow(data.values, extent=(bounds), **im_kwargs)
if colorbar:
if colorbar not in ['vertical', 'horizontal']:
colorbar = 'vertical'
self.cbar = add_colorbar(self.im, orientation=colorbar)
else:
self.cbar = None
self.data = data
def Contour(x,
y,
z,
cmap="bwr",
filename=None,
ax=None,
colorbar="vertical",
ax_kwargs={}):
ax = get_ax(ax)
ax.set_aspect("auto")
im = ax.contourf(x, y, z, cmap=cmap)
if colorbar:
if colorbar not in ["vertical", "horizontal"]:
colorbar = default_colorbar_orientation
add_colorbar(im, colorbar)
ax_formatting(ax, **ax_kwargs)
return im
def Contour(x,
y,
z,
cmap='bwr',
filename=None,
ax=None,
colorbar='vertical',
ax_kwargs={}):
ax = get_ax(ax)
ax.set_aspect('auto')
im = ax.contourf(x, y, z, cmap=cmap)
if colorbar:
if colorbar not in ['vertical', 'horizontal']:
colorbar = default_colorbar_orientation
add_colorbar(im, colorbar)
ax_formatting(ax, **ax_kwargs)
return im
def plot_window(self, ax=None):
ax = get_ax(ax)
self.window.plot(ax=ax)
def Histogram(
data,
bins=10,
range=None,
density=False,
proportion=False,
multiplier=1,
invert=False,
engine="matplotlib",
ax=None,
params={},
):
""" A convenience function for generating histograms
Parameters
----------
data: array_like
Input data (see np.histogram)
bins: int or sequence of scalars or str, optional; default = 10
Defines histrogram bins (see np.histogram)
range: (float, float), optional; default = None
Lower and upper bounds of the bins (see np.histogram)
density: bool, optional; default = False
number of samples if False, density if True (see np.histogram)
density and proportion cannot both be True
density is such that the area under the histogram is 1
np.sum(frequencies * np.diff(edges)) == 1
proportion: bool, optional; default = False
number of samples if False, density if True (see np.histogram)
density and proportion cannot both be True
proportion is such that the sum of bar heights is 1
np.sum(frequencies) == 1
if proportion is a numeric value, normalized to that instead
multiplier: numeric, optional; default = 1
A scalar to multiple all the counts by
invert: bool, optional; default = False
If True, all frequencies are multiplied by -1 to flip upside down
engine: str, optional; default = 'holoviews'
Which plotting engine is to be used. Defaults to holoviews
Currently supported: ['holoviews']
Returns
-------
if 'engine' is 'holoviews', returns hv.Histogram
if 'engine' is 'matplotlib', returns ax
"""
data = np.asarray(data)
if proportion and density:
raise ValueError("proportion and density cannot both be enabled")
weights = (np.ones_like(data) * proportion /
data.size if proportion else np.ones_like(data))
if invert:
weights *= -1
weights = weights * multiplier
frequencies, edges = np.histogram(data,
bins,
range,
density=density,
weights=weights)
if engine == "holoviews":
if invert:
frequencies *= -1
frequencies = frequencies * multiplier
if proportion or density:
raise NotImplementedError
hist = _histogram_holoviews(edges, frequencies)
return hist
elif engine == "matplotlib":
ax = get_ax(ax)
if params.get("histtype") == "line":
params.pop("histtype")
cumulative = params.pop("cumulative", False)
counts, edges = np.histogram(data,
bins=bins,
range=range,
density=density,
weights=weights)
if cumulative:
counts = np.cumsum(counts)
align = params.pop("align", "mid")
if align == "left":
bin_align_points = edges[:-1]
elif align == "right":
bin_align_points = edges[1:]
else:
bin_align_points = np.mean([edges[:-1], edges[1:]], axis=0)
orientation = params.pop("orientation", "vertical")
if orientation == "vertical":
ax.plot(bin_align_points, counts, **params)
elif orientation == "horizontal":
ax.plot(counts, bin_align_points, **params)
else:
ax.hist(data,
bins=bins,
range=range,
density=density,
weights=weights,
**params)
return ax
else:
raise ValueError(
f"Engine not implemented: {engine}. Choose one of: ['holoviews','matplotlib']"
)
def Raster(data, ax=None, eventplot_params={}, **kwargs):
ax = get_ax(ax)
eventplot_params = ChainMap(eventplot_params,
default_raster_eventplot_params)
ax.eventplot(data, **eventplot_params)
return ax