def plot_scatter(filename=None, xs=None, ys=None, size=20, pch='x', colour='white', title=None, skipheader=False):
"""plot (X,y) pairs from columns in a file or 2 separate files
Arguments:
xs List of values to use on X axis [default: None]
ys List of values to use on y axis [default: None]
filename Name of a csv file [default: None]
skipheader Skip the first row [default: False]
size Marketsize [default: 20]
pch Markershape [default: x]
colour Marker colour [default: white]
title Title [default: ]
"""
if filename:
data = helpers.read_xy_pairs_from_csv(filename, X=0, y=1, header=skipheader)
xs = [i[0] for i in data]
ys = [i[1] for i in data]
else:
xs = [line.strip() for line in open(xs, 'r').readlines()]
ys = [line.strip() for line in open(ys, 'r').readlines()]
if skipheader:
xs = xs[1:]
ys = ys[1:]
xs = map(float, xs)
ys = map(float, ys)
if (xs is None or ys is None):
raise Exception("Missing `xs` or `ys` data.")
plotted = set()
if title:
print(helpers.box_text(title, 2 * len(get_scale(xs, False, size)) + 1))
print("-" * (2 * len(get_scale(xs, False, size)) + 2))
for y in get_scale(ys, True, size):
print("|", end=' ')
for x in get_scale(xs, False, size):
point = " "
for (i, (xp, yp)) in enumerate(zip(xs, ys)):
if xp <= x and yp >= y and (xp, yp) not in plotted:
point = pch
#point = str(i)
plotted.add((xp, yp))
if x == 0 and y == 0:
point = "o"
elif x == 0:
point = "|"
elif y == 0:
point = "-"
helpers.printcolour(point, True, colour)
print("|")
print("-" * (2 * len(get_scale(xs, False, size)) + 2))
def plot_hist(data, height=20.0, bincount=None, pch="o", colour="white", title="", xlab=None, showSummary=False):
"""make a histogram for continuous variable.
Arguments:
data: List of numbers or file with numbers
height: The height of the histogram in # of lines
bincount: Number of bins in the histogram
pch: Shape of the bars in the plot
colour: Colour of the bars in the terminal
title: Title at the top of the plot
xlab: Boolen value for whether or not to display x-axis labels
showSummary: Boolean value for whether or not to display a summary
"""
if pch is None:
pch = "o"
colour = helpers.get_colour(colour)
min_val, max_val = None, None
n, mean = 0., 0.
for number in read_numbers(data):
n += 1
if not min_val or number < min_val:
min_val = number
if not max_val or number > max_val:
max_val = number
mean += number
mean /= n
bins = list(calc_bins(n, min_val, max_val, bincount))
hist = {}
for i in range(len(bins)):
hist[i] = 0
for number in read_numbers(data):
for i, b in enumerate(bins):
if number < b:
hist[i] += 1
break
min_y, max_y = min(hist.values()), max(hist.values())
ys = list(helpers.drange(min_y, max_y, (max_y-min_y)/height))
ys.reverse()
nlen = max(len(str(min_y)), len(str(max_y))) + 1
if title:
print helpers.box_text(title, len(hist)*2, nlen)
print
used_labs = set()
for y in ys:
ylab = str(int(y))
if ylab in used_labs:
ylab = ""
else:
used_labs.add(ylab)
ylab = " "*(nlen - len(ylab)) + ylab + "|"
print ylab,
for i in range(len(hist)):
if y < hist[i]:
helpers.printcolor(pch, True, colour)
else:
helpers.printcolor(" ", True, colour)
print
xs = hist.keys() * 2
print " "*(nlen+1) + "-"*len(xs)
if xlab:
for i in range(0, nlen):
helpers.printcolor(" "*(nlen+1), True, colour)
for x in range(0, len(hist)):
num = str(bins[x])
if x%2==0:
print " ",
elif i < len(num):
print num[i],
print
center = max(map(len, map(str, [n, min_val, mean, max_val])))
center += 15
if showSummary:
print
print "-"*(2 + center)
print "|" + "Summary".center(center) + "|"
print "-"*(2 + center)
summary = "|" + ("observations: %d" % n).center(center) + "|\n"
summary += "|" + ("min value: %f" % min_val).center(center) + "|\n"
summary += "|" + ("mean : %f" % mean).center(center) + "|\n"
summary += "|" + ("max value: %f" % max_val).center(center) + "|\n"
summary += "-"*(2 + center)
print summary
