PyVF/tpp_analyze_pt2.py at master · constructor-s/PyVF · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
import json
import time
from matplotlib.ticker import ScalarFormatter, NullFormatter

import pyvf.strategy.Model
import pyvf.plot
import pandas as pd
import numpy as np
import scipy.stats
import matplotlib as mpl
import matplotlib.pyplot as plt
from mpl_toolkits.axes_grid1.inset_locator import inset_axes
from datetime import datetime
from pytz import timezone
from collections import namedtuple
from pathlib import Path
from argparse import ArgumentParser
from pandas.api.types import CategoricalDtype

parser = ArgumentParser()
parser.add_argument("-i", "--input", required=True, type=str, help="Output from part 1")
parser.add_argument("-o", "--output-folder", required=True, type=str, help="Figure export folder")
args = parser.parse_args()

df = pd.read_csv(args.input, parse_dates=["timestamp"])  # type: pd.DataFrame

tests_start = df[df["comment"]!="invalid"].groupby(["id", "eye"])["timestamp"].min()
tests_end = df[df["comment"]!="invalid"].groupby(["id", "eye"])["timestamp"].max()
tests_duration = tests_end - tests_start
tests_duration_weeks = tests_duration.apply(lambda x: x.days/7.0)
tests_n = df[df["comment"]!="invalid"].groupby(["id", "eye"])["timestamp"].count()

#%%
for name, group_all in df.groupby(["id", "eye"]):
    # TODO: This is to plot only one subject, make this an argument
    # if name[0] not in [11]:
    #     continue
    print(name)
    print(group_all.loc[:, ['timestamp', 'fl_error', 'fl_total', 'fp_error', 'fp_total', 'fn_error', 'fn_total',
       'md', 'psd', 'vfi', 'ght']].to_string())
    mask = ((group_all["comment"]!="exclude") &
              (group_all["comment"] != "invalid") &
              (group_all["routine"] == "TPP242") &
              (group_all["fl_error"] / group_all["fl_total"] <= 0.20) &
              (group_all["fp_error"] / group_all["fp_total"] <= 0.20) &
              (group_all["fn_error"] / group_all["fn_total"] <= 0.20))
    mask.iloc[-1:] = True # Force include last test
    # print(mask)
    # mask &= group_all["timestamp"] > datetime(2021, 1, 1)
    group = group_all[mask]

    fig, ax = plt.subplots(1, 1, figsize=(8.5, 8.5))
    plotter = pyvf.plot.VFPlotManager()
    if name[1].upper() == "OD":
        plotter.pattern = pyvf.strategy.PATTERN_P24D2
    elif name[1].upper() == "OS":
        plotter.pattern = pyvf.strategy.PATTERN_P24D2_OS
    else:
        raise ValueError(f"{name} is invalid")
    plotter.ax = ax
    plotter.create_axes()

    plotter.ax.set_title("-".join(map(str, name)))
    plotter.ax.xaxis.set_minor_formatter(NullFormatter())
    plotter.ax.yaxis.set_minor_formatter(NullFormatter())
    for i, axin in enumerate(plotter.axins):
        if i in (25, 34):
            continue
        ax2 = axin.twiny()
        ax2.hist(group[f"L{i}"], bins=np.arange(-4, 40.1, 4), orientation='horizontal', color=plt.cm.gray(mpl.colors.Normalize(-35, 45)(group[f"L{i}"].mean())), zorder=-10)
        ax2.set(xticks=())
        ax2.set_ylim([-1, 40])
        axin.scatter(group["timestamp"], group[f"L{i}"], s=2, c=group[f"L{i}"], cmap=plt.cm.viridis, norm=mpl.colors.Normalize(0, 60), zorder=0)
        lfit = np.poly1d(np.polyfit(pd.to_numeric(group["timestamp"]), group[f"L{i}"], 1))
        axin.plot(group["timestamp"], lfit(pd.to_numeric(group["timestamp"])), 'k--', linewidth=1, zorder=10)
        axin.set_ylim([-1, 40])

        regression_result = scipy.stats.linregress((group["timestamp"]-group["timestamp"].iloc[0]).dt.days/365.25, group[f"L{i}"])
        if regression_result.slope < 0 and regression_result.pvalue < 0.05:
            legend = axin.legend([], facecolor="None", edgecolor="red",
                                 title=f"{regression_result.slope:+.1f},p={regression_result.pvalue:.3f}")
        else:
            legend = axin.legend([], facecolor="None", edgecolor="None",
                                 title=f"{regression_result.slope:+.1f},p={regression_result.pvalue:.3f}")
        plt.setp(legend.get_title(), fontsize='xx-small')

    if name[1].upper() == "OD":
        top_left_i = 0
    elif name[1].upper() == "OS":
        top_left_i = 3
    else:
        raise ValueError(f"{name} is invalid")
    plotter.axins[top_left_i].tick_params(labelleft=True, labeltop=True)
    plotter.axins[top_left_i].set(xticks=group["timestamp"].iloc[[0, -1]], yticks=np.arange(0, 40.1, 10))
    plotter.axins[top_left_i].set_xlabel("Threshold (dB)")
    plt.setp(plotter.axins[top_left_i].get_xticklabels(), rotation=90)

    axi = inset_axes(plotter.ax, width="100%", height="100%", loc=3,
                     bbox_to_anchor=(0.03, 0.02, 0.26, 0.17), bbox_transform=plotter.ax.transAxes,
                     borderpad=0, axes_kwargs={"zorder": 10})
    axi.scatter(group["timestamp"], group["md"], s=2, c=group["md"], marker="s", cmap=plt.cm.viridis, norm=mpl.colors.Normalize(-30, 30), zorder=0)
    regression_result = scipy.stats.linregress((group["timestamp"] - group["timestamp"].iloc[0]).dt.days / 365.25, group["md"])
    if regression_result.slope < 0 and regression_result.pvalue < 0.05:
        axi.legend([], facecolor="None", edgecolor="red", title=f"MD:{regression_result.slope:+.3f} db/yr,p={regression_result.pvalue:.3f}")
    else:
        axi.legend([], facecolor="None", edgecolor="None", title=f"MD:{regression_result.slope:+.3f} db/yr,p={regression_result.pvalue:.3f}")
    lfit = np.poly1d(np.polyfit(pd.to_numeric(group["timestamp"]), group["md"], 1))
    axi.plot(group["timestamp"], lfit(pd.to_numeric(group["timestamp"])), 'k--', linewidth=1, zorder=+20)
    ylim = axi.get_ylim()
    axi.set_xticks(group["timestamp"].iloc[[0, -1]])
    plt.setp(axi.get_xticklabels(), rotation=90)
    ax2 = axi.twiny()
    ax2.hist(group["md"],
             bins=np.arange(min(-30, np.floor(group["md"].min())), max(0, np.ceil(group["md"].max()))+2, 1.0),
             orientation='horizontal', color=plt.cm.gray(mpl.colors.Normalize(-45, 15)(group["md"].mean())),
             zorder=-10)
    ax2.set(xticks=())
    ax2.grid(axis="y", linestyle="dashed")

    axi = inset_axes(plotter.ax, width="100%", height="100%", loc=3,
                     bbox_to_anchor=(0.73, 0.02, 0.26, 0.17), bbox_transform=plotter.ax.transAxes,
                     borderpad=0, axes_kwargs={"zorder": 10})
    axi.scatter(group["timestamp"], group["psd"], s=2, c=group["psd"], marker="s", cmap=plt.cm.viridis_r,
                norm=mpl.colors.Normalize(-20, 20), zorder=0)
    regression_result = scipy.stats.linregress((group["timestamp"] - group["timestamp"].iloc[0]).dt.days / 365.25, group["psd"])
    axi.legend([], facecolor="None", edgecolor="None", title=f"PSD:{regression_result.slope:+.3f} db/yr,p={regression_result.pvalue:.3f}")
    lfit = np.poly1d(np.polyfit(pd.to_numeric(group["timestamp"]), group["psd"], 1))
    axi.plot(group["timestamp"], lfit(pd.to_numeric(group["timestamp"])), 'k--', linewidth=1, zorder=+20)
    ylim = axi.get_ylim()
    axi.set_xticks(group["timestamp"].iloc[[0, -1]])
    plt.setp(axi.get_xticklabels(), rotation=90)
    ax2 = axi.twiny()
    ax2.hist(group["psd"],
             bins=np.arange(min(0, np.floor(group["psd"].min())), max(20, np.ceil(group["psd"].max())) + 2, 1.0),
             orientation='horizontal', color=plt.cm.gray_r(mpl.colors.Normalize(-10, 30)(group["psd"].mean())),
             zorder=-10)
    ax2.set(xticks=())
    ax2.grid(axis="y", linestyle="dashed")

    fig.show()
    fig.savefig(Path(args.output_folder) / ("-".join(map(str, name))+".pdf"))
    # fig.savefig(Path(args.output_folder) / ("-".join(map(str, name))+".png"), dpi=300)
    plt.close(fig)

    # break