Commit 011a6cb
Director of Analytics
updated w/ new python histogram plotter
1 parent 06bef79 commit 011a6cb
2 files changed
Lines changed: 108 additions & 0 deletions
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
@@ -0,0 +1,2 @@ | |||
| 1 | + *.csv | ||
| 2 | + *.txt | ||
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
@@ -0,0 +1,106 @@ | |||
| 1 | + import math | ||
| 2 | + from collections import Counter | ||
| 3 | + import optparse | ||
| 4 | + | ||
| 5 | + | ||
| 6 | + def drange(start, stop, step=1.0): | ||
| 7 | + "generate between 2 numbers w/ optional step" | ||
| 8 | + r = start | ||
| 9 | + while r < stop: | ||
| 10 | + yield r | ||
| 11 | + r += step | ||
| 12 | + | ||
| 13 | + def calc_bins(n, min_val, max_val, h=None): | ||
| 14 | + "calculate number of bins for the histogram" | ||
| 15 | + if not h: | ||
| 16 | + h = math.log(n + 1, 2) | ||
| 17 | + bin_width = (max_val - min_val) / h | ||
| 18 | + for b in drange(min_val, max_val, bin_width): | ||
| 19 | + yield b | ||
| 20 | + | ||
| 21 | + def plot_hist(f, height=20, bincount=None): | ||
| 22 | + "plot a histogram given a file of numbers" | ||
| 23 | + #first apss | ||
| 24 | + min_val, max_val = None, None | ||
| 25 | + n = 0. | ||
| 26 | + for number in open(f): | ||
| 27 | + n += 1 | ||
| 28 | + number = float(number) | ||
| 29 | + | ||
| 30 | + if not min_val or number < min_val: | ||
| 31 | + min_val = number | ||
| 32 | + if not max_val or number > max_val: | ||
| 33 | + max_val = number | ||
| 34 | + | ||
| 35 | + bins = list(calc_bins(n, min_val, max_val, bincount)) | ||
| 36 | + hist = Counter() | ||
| 37 | + for number in open(f): | ||
| 38 | + number = float(number) | ||
| 39 | + for i, b in enumerate(bins): | ||
| 40 | + if number < b: | ||
| 41 | + hist[i-1] += 1 | ||
| 42 | + # print "breaking" | ||
| 43 | + break | ||
| 44 | + | ||
| 45 | + | ||
| 46 | + | ||
| 47 | + min_y, max_y = min(hist.values()), max(hist.values()) | ||
| 48 | + | ||
| 49 | + ys = list(drange(min_y, max_y, (max_y-min_y)/height)) | ||
| 50 | + ys.reverse() | ||
| 51 | + | ||
| 52 | + nlen = max(len(str(min_y)), len(str(max_y))) + 1 | ||
| 53 | + | ||
| 54 | + | ||
| 55 | + for y in ys: | ||
| 56 | + ylab = str(y) | ||
| 57 | + ylab += " "*(nlen - len(ylab)) + "|" | ||
| 58 | + | ||
| 59 | + print ylab, | ||
| 60 | + | ||
| 61 | + for i in range(len(hist)): | ||
| 62 | + if y < hist[i]: | ||
| 63 | + print "o", | ||
| 64 | + else: | ||
| 65 | + print " ", | ||
| 66 | + print | ||
| 67 | + xs = hist.keys() * 2 | ||
| 68 | + | ||
| 69 | + print " "*(nlen+1) + "-"*len(xs) | ||
| 70 | + | ||
| 71 | + for i in range(0, nlen): | ||
| 72 | + print " "*(nlen+1), | ||
| 73 | + for x in range(0, len(hist)): | ||
| 74 | + n = str(bins[x]) | ||
| 75 | + if x%2==0: | ||
| 76 | + print " ", | ||
| 77 | + elif i < len(n): | ||
| 78 | + print n[i], | ||
| 79 | + print | ||
| 80 | + | ||
| 81 | + summary = "Summary\n--------\nMax: %s\nMin:%s" % (min_val, max_val) | ||
| 82 | + print summary | ||
| 83 | + | ||
| 84 | + | ||
| 85 | + if __name__=="__main__": | ||
| 86 | + | ||
| 87 | + parser = optparse.OptionParser() | ||
| 88 | + parser.add_option('-f', '--file', help='a file containing a column of numbers', | ||
| 89 | + default=None, dest='f') | ||
| 90 | + parser.add_option('-b', '--bins', help='number of bins in the histogram', | ||
| 91 | + default=None, dest='b') | ||
| 92 | + parser.add_option('-s', help='height of the histogram (in lines)', | ||
| 93 | + default=20, dest='h') | ||
| 94 | + | ||
| 95 | + (opts, args) = parser.parse_args() | ||
| 96 | + | ||
| 97 | + if opts.f is None: | ||
| 98 | + opts.f = args[0] | ||
| 99 | + | ||
| 100 | + plot_hist(opts.f, opts.h, opts.b) | ||
| 101 | + | ||
| 102 | + | ||
| 103 | + | ||
| 104 | + | ||
| 105 | + | ||
| 106 | + | ||
0 commit comments