def plot_prim_time_series(boxes, data, outcome):
x, results = data
outcomes = results[outcome]
try:
time = results.get('TIME')[0, :]
except KeyError:
time = np.arange(0, outcomes.shape[1])
#the plotting
grid = gridspec.GridSpec(len(boxes)-1, 1)
grid.update(wspace = 0.05,
hspace = 0.4)
figure = plt.figure()
for i, box in enumerate(boxes[0:-1]):
logical = in_box(x, box.box)
#make the axes
ax = figure.add_subplot(grid[i, 0])
value = outcomes[logical]
ax.plot(time.T[:, np.newaxis], value.T)
return figure
def make_boxes(tree, data, classify, threshold):
'''
Function that turns a classification tree into prim boxes, including
the scenario discovery metrics.
:param tree: the return from :func:`orangeFunctions.tree`.
:param data: the return from :meth:`perform_experiments`.
:param classify: the classify function used in making the tree.
:param threshold: the minimum mean that the boxes should meet.
:return: a list of prim boxes.
'''
branches = find_branches(tree.tree, [("root", "")])
data, results = data
init_box = make_box(data)
boxes = []
for branch in branches:
box = copy.copy(init_box)
for name in branch[1:]:
name, limit = name
if name=="root" or type(name) != StringType:
continue
if limit.startswith('>'):
limit = limit[1:]
limit = float(limit)
box[name][0] = limit
else:
limit = limit[2:]
limit = float(limit)
box[name][1] = limit
boxes.append(box)
y = classify(results)
n = y.shape[0]
threshold=0.8
threshold_type=1
#turn boxes into prim objects
new_boxes = []
for box in boxes:
logical = in_box(data, box)
box = Prim(data[logical], y[logical], box, y[logical].shape[0]/n)
new_boxes.append(box)
boxes = prim_hdr(new_boxes, threshold=threshold, threshold_type=threshold_type)
boxes = calculate_sd_metrics(boxes, y, threshold, threshold_type)
return boxes
