def heatmap(df, cmap, norm=None, col_ranker=None, idx_ranker=None, figsize=None):
"""Plot a matrix heatmap"""
print(utils.describe_df(df, flatten=True))
if col_ranker is not None:
ranks = [col_ranker(df.loc[:, c]) for c in df.columns]
columns, _ = zip(*sorted(zip(df.columns, ranks), key=lambda x: x[1]))
df = df[list(columns)]
if idx_ranker is not None:
ranks = [idx_ranker(df.loc[i, :]) for i in df.index]
index, _ = zip(*sorted(zip(df.index, ranks), key=lambda x: x[1]))
df = df.reindex(list(index))
plt.close('all')
if figsize is None:
figsize = (len(df.columns) * 0.5 + 0.5, len(df.index) * 0.48)
fig, ax = plt.subplots(figsize=figsize)
im = ax.pcolor(df, cmap=cmap, norm=norm, vmin=df.min().min(), vmax=df.max().max())
x = np.arange(len(df.columns)) + 0.5
y = np.arange(len(df.index)) + 0.5
min_idx, min_col, max_idx, max_col = unravel_index(df)
min_x, min_y, max_x, max_y = x[min_col], y[min_idx], x[max_col], y[max_idx]
ax.plot(min_x, min_y, marker='x', markersize=10, color='black')
ax.plot(max_x, max_y, marker='x', markersize=10, color='black')
ax.set_xticks(x, minor=False)
ax.set_yticks(y, minor=False)
ax.set_xticklabels(df.columns, minor=False)
ax.set_yticklabels(df.index, minor=False)
# want a more natural, table-like display
ax.invert_yaxis()
ax.xaxis.tick_top()
plt.xticks(rotation=45)
plt.grid(False)
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size=0.35, pad=0.15)
plt.colorbar(im, cax=cax)
# Turn off all the ticks
ax = plt.gca()
for t in ax.xaxis.get_major_ticks():
t.tick1On = False
t.tick2On = False
for t in ax.yaxis.get_major_ticks():
t.tick1On = False
t.tick2On = False
plt.show()
def evolution(df, cmap, norm=None, rank=None, sentiment=lambda sr: sr.mean(), figsize=None):
"""Combine multiple time series into a heatmap and plot"""
print(utils.describe_df(df, flatten=True))
if rank is not None:
if isinstance(rank, str):
if rank == 'correlation':
rank = lambda sr: -np.corrcoef(df[df.columns[0]], sr)[0, 1]
elif rank == 'last':
rank = lambda sr: -sr.iloc[-1]
elif rank == 'mean':
rank = lambda sr: -sr.mean()
ranks = [rank(df[c]) for c in df.columns]
columns, _ = zip(*sorted(zip(df.columns, ranks), key=lambda x: x[1]))
df = df[list(columns)]
index = trunk_dt_index(df.index)
columns = df.columns
plt.close('all')
if figsize is None:
figsize = (14, len(columns) * 0.45)
fig, ax = plt.subplots(figsize=figsize)
im = ax.pcolor(df.transpose(), cmap=cmap, norm=norm, vmin=df.min().min(), vmax=df.max().max())
x = np.arange(len(index)) + 0.5
y = np.arange(len(columns)) + 0.5
min_x, min_y, max_x, max_y = unravel_index(df)
min_x, min_y, max_x, max_y = x[min_x], y[min_y], x[max_x], y[max_y]
ax.plot(min_x, min_y, marker='x', markersize=10, color='black')
ax.plot(max_x, max_y, marker='x', markersize=10, color='black')
# xticks
nticks = 6
tick_interval = math.ceil(len(x) / nticks)
ax.set_xticks(x[::tick_interval], minor=False)
ax.set_xticklabels(index[::tick_interval], minor=False)
# yticks
ax.set_yticks(y, minor=False)
ax.set_yticklabels(columns, minor=False)
ax.invert_yaxis()
divider = make_axes_locatable(ax)
cax_right = divider.append_axes("right", size=0.35, pad=0.15)
plt.colorbar(im, cax=cax_right)
cax_top = divider.append_axes("top", size=0.35, pad=0.15)
sentiment_sr = df.apply(sentiment, axis=1)
cax_top.pcolor([sentiment_sr], cmap=cmap, norm=norm, vmin=df.min().min(), vmax=df.max().max())
cax_top.set_yticks([0.5], minor=False)
cax_top.set_yticklabels(["sentiment"], minor=False)
plt.setp(cax_top.get_xticklabels(), visible=False)
plt.grid(False)
# Turn off all the ticks
ax = plt.gca()
for t in ax.xaxis.get_major_ticks():
t.tick1On = False
t.tick2On = False
for t in ax.yaxis.get_major_ticks():
t.tick1On = False
t.tick2On = False
plt.show()
def timesr_matrix(df, bands=None, rank=None, ncols=3, figsize=None):
"""Plot group of line charts"""
print(utils.describe_df(df, flatten=True))
nsubplots = len(df.columns)
nrows = math.ceil(nsubplots / ncols)
if rank is not None:
ranks = [rank(df[c]) for c in df.columns]
columns, _ = zip(*sorted(zip(df.columns, ranks), key=lambda x: x[1]))
else:
columns = df.columns
plt.close('all')
if figsize is None:
figsize = (4 * ncols, 2.5 * nrows)
fig, axes = plt.subplots(nrows=nrows, ncols=ncols, figsize=figsize)
for i, ax in enumerate(axes.flat):
if i < nsubplots:
sr = df[columns[i]].copy()
sr.index = trunk_dt_index(sr.index)
x = list(range(len(sr.index)))
y = sr.bfill().values
min_x, max_x, min_y, max_y = np.min(x), np.max(x), np.min(
y), np.max(y)
offset = 0.2 * (max_y - min_y)
min_y -= offset
max_y += offset
gradient_fill(ax, x, y, 'grey', 0.5, ylim=(min_y, max_y))
if bands is not None:
upper_band = bands[0][columns[i]]
lower_band = bands[1][columns[i]]
mean = (lower_band + upper_band) / 2
ax.fill_between(x, upper_band, mean, color='green', alpha=0.2)
ax.fill_between(x,
mean,
lower_band,
color='orangered',
alpha=0.2)
ax.plot(sr.values.argmin(),
sr.min(),
marker='x',
markersize=10,
color='black')
ax.plot(sr.values.argmax(),
sr.max(),
marker='x',
markersize=10,
color='black')
ax.set_xticks([sr.values.argmin(), sr.values.argmax()])
ax.set_xticklabels([sr.argmin(), sr.argmax()])
ax.ticklabel_format(style='sci', axis='y', scilimits=(0, 0))
ax.set_xlim((min_x, max_x))
ax.set_ylim((min_y, max_y))
ax.set_title(columns[i])
ax.grid(False)
else:
fig.delaxes(ax)
plt.tight_layout()
plt.show()
def depth(orderbooks, colors=None, rank=None, ncols=3, figsize=None):
"""Plot depth graphs from order books"""
df = pd.DataFrame(np.array(list(orderbooks.values())).flatten())
print(utils.describe_df(df))
pairs = list(orderbooks.keys())
nsubplots = len(orderbooks.keys())
nrows = math.ceil(nsubplots / ncols)
if rank is not None:
ranks = [rank(orderbooks[p]) for p in pairs]
pairs, _ = zip(*sorted(zip(pairs, ranks), key=lambda x: x[1]))
plt.close('all')
if figsize is None:
figsize = (4 * ncols, 2.5 * nrows)
fig, axes = plt.subplots(nrows=nrows, ncols=ncols, figsize=figsize)
if colors is None:
colors = ('grey', 'grey')
for i, ax in enumerate(axes.flat):
if (i < nsubplots):
orderbook = orderbooks[pairs[i]]
bids = orderbook[orderbook > 0]
asks = np.abs(orderbook[orderbook < 0])
split = len(bids.index)
x = orderbook.index
y = np.abs(orderbook.values)
min_x = min(x)
max_x = max(x)
min_y = 0
max_y = max(y)
c1, c2 = colors
gradient_fill(ax,
x[:split],
bids.values,
c1,
0.5,
ylim=(min_y, max_y))
gradient_fill(ax,
x[split:],
asks.values,
c2,
0.5,
ylim=(min_y, max_y))
ax.set_xlim((min_x, max_x))
ax.set_ylim((min_y, max_y))
ax.set_xticks([x[0], x[split], x[-1]])
ax.ticklabel_format(style='sci', axis='both', scilimits=(0, 0))
ax.set_title(pairs[i])
ax.grid(False)
else:
fig.delaxes(ax)
plt.tight_layout()
plt.show()
def hist_matrix(df,
cmap,
bins=20,
norm=None,
rank=None,
axvlines=None,
ncols=3,
figsize=None):
"""Plot group of histograms"""
print(utils.describe_df(df, flatten=True))
nsubplots = len(df.columns)
nrows = math.ceil(nsubplots / ncols)
if rank is not None:
ranks = [rank(df[c]) for c in df.columns]
columns, _ = zip(*sorted(zip(df.columns, ranks), key=lambda x: x[1]))
else:
columns = df.columns
plt.close('all')
if figsize is None:
figsize = (4 * ncols, 2.5 * nrows)
fig, axes = plt.subplots(nrows=nrows, ncols=ncols, figsize=figsize)
for i, ax in enumerate(axes.flat):
if i < nsubplots:
sr = df[columns[i]].copy()
hist, _bins = np.histogram(sr, bins=bins)
width = np.diff(_bins)
center = (_bins[:-1] + _bins[1:]) / 2
if cmap is None:
colors = 'grey'
else:
if norm is None:
norm = plt.Normalize()
colors = cmap(norm(_bins))
ax.bar(center,
hist,
color=colors,
align='center',
width=width,
alpha=0.5)
if axvlines is None:
ax.axvline(sr.mean(), color='black', lw=1)
else:
for axvline in axvlines:
if callable(axvline):
ax.axvline(axvline(sr), color='black', lw=1)
else:
ax.axvline(axvline, color='black', lw=1)
ax.set_title(columns[i])
ax.grid(False)
else:
fig.delaxes(ax)
plt.tight_layout()
plt.show()