Python3 matplotlib script that plots histograms for features of an ML corpus

features-histogram.py

#!/usr/bin/env python3
"""
usage: {} < corpus.json > plot.pdf

The corpus file must contain one JSON document per line,
features must be stored in a field names "{}",
classes in a field names "{}".
"""

import sys
import json
from collections import defaultdict
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.ticker import EngFormatter

FEATURES_FIELD = "features"
CLASS_FIELD = "category"
RATIO = (2, 3)


# display help if no data on stdin
if sys.stdin.isatty():
    print(__doc__.format(sys.argv[0], FEATURES_FIELD, CLASS_FIELD))
    sys.exit(1)


# load corpus from the corpus file
features = []
class_features = defaultdict(list)
for line in sys.stdin:
    data = json.loads(line)
    features.append(data[FEATURES_FIELD])
    class_features[data[CLASS_FIELD]].append(data[FEATURES_FIELD])
features = np.transpose(np.array(features))
class_features = {
    cl: np.transpose(np.array(fts))
    for cl, fts in class_features.items()}

# setup global figure parameters
plotsize = (len(features), len(class_features) + 1)
fig = plt.figure(figsize=tuple(x * plotsize[i] for i, x in enumerate(RATIO)))
plt.rcParams["axes.grid"] = True
plt.rcParams["grid.linestyle"] = "dotted"

# display information about the corpus as suptitle
desc = "samples:{} (".format(len(features[0]))
desc += ", ".join([
    "{}:{}".format(cl, len(fts[0]))
    for cl, fts in class_features.items()])
fig.suptitle(desc + ")")

# display features subplots
ylims = []
for i, vals in enumerate(features):
    axis = plt.subplot2grid(plotsize, (i, 0))
    axis.yaxis.set_major_formatter(EngFormatter())
    if i == 0:
        plt.title("all")
    axis.set_ylabel("feature #{}".format(i))
    axis.get_yaxis().set_label_coords(-0.15, 0.5)
    plt.hist(vals)
    ylims.append(axis.get_ylim())

# display per-class features subplots
for i, c in enumerate(class_features):
    for j, vals in enumerate(class_features[c]):
        axis = plt.subplot2grid(plotsize, (j, i+1))
        axis.yaxis.set_major_formatter(EngFormatter())
        if j == 0:
            plt.title("{}".format(c))
        plt.hist(vals)
        axis.set_ylim(ylims[j])

# fix layout
plt.tight_layout()
fig.subplots_adjust(top=0.90)

# output the resulting plot
plt.savefig(sys.stdout.buffer, format="pdf")
plt.close()