DSensDp/evaluate.py at main · fmenat/DSensDp · 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
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
import yaml
import argparse
import gc
import os
from pathlib import Path
from tqdm import tqdm
import numpy as np
import pandas as pd

from code.datasets.utils import load_structure
from code.evaluate.utils import load_data_per_fold, save_results, gt_mask
from code.metrics.metrics import ClassificationMetrics, SoftClassificationMetrics

def classification_metric(
                preds_p_run,
                indexs_p_run,
                data_ground_truth,
                ind_save,
                dir_folder = "",
                task_type="classification",
                ):
    TARGET_NAMES = data_ground_truth.target_names
    R = len(preds_p_run)

    df_runs = []
    df_runs_diss = []
    df_per_run_fold = []
    y_true_concatenated = []
    y_pred_cate_concatenated = [] #for classification
    for r in tqdm(range(R)):
        indexs_p_run_r = indexs_p_run[r]
        preds_p_run_r = preds_p_run[r]
        df_per_fold = []
        for f in tqdm(range(len(indexs_p_run_r))):
            y_true, y_pred = gt_mask(data_ground_truth, indexs_p_run_r[f]), preds_p_run_r[f]
            y_true = np.squeeze(y_true)
            y_pred = np.squeeze(y_pred)

            y_true_concatenated.append(y_true)

            if task_type == "classification":
                y_pred_cate = np.argmax(y_pred, axis = -1).astype(np.uint8)
                y_pred_cate_concatenated.append(y_pred_cate)

                d_me = ClassificationMetrics()
                dic_res = d_me(y_pred_cate, y_true)

                d_me_aux = SoftClassificationMetrics(["ENTROPY","LOGP", "PMAX", "PTRUE"]+(["AUC"] if y_true.max() ==1 else []))
                dic_res.update(d_me_aux(y_pred, y_true))

                d_me = ClassificationMetrics(["F1 none", "R none", "P none", "ntrue", 'npred'])
                dic_des = d_me(y_pred_cate, y_true)
                df_des = pd.DataFrame(dic_des)
                df_des.index = TARGET_NAMES if type(TARGET_NAMES) == list else ["negative", "positive"]
                df_runs_diss.append(df_des)

            elif task_type == "multilabel":
                y_pred_cate = (y_pred>= 0.5).astype(np.uint8)
                y_pred_cate_concatenated.append(y_pred_cate)

                d_me = ClassificationMetrics(["F1 macro", "R macro", "P macro",
                                              "F1 weighted", "R weighted", "P weighted",
                                              "F1 micro", "R micro", "P micro", "ENTROPY"], "multilabel")
                dic_res = d_me(y_pred_cate, y_true)

                d_me_aux = SoftClassificationMetrics(["mAP", "ENTROPY"])
                dic_res.update(d_me_aux(y_pred, y_true))

                d_me = ClassificationMetrics(["F1 none", "R none", "P none", "ntrue", 'npred'], "multilabel")
                dic_des = d_me(y_pred_cate, y_true)
                df_des = pd.DataFrame(dic_des)
                df_des.index = TARGET_NAMES
                df_runs_diss.append(df_des)


            df_res = pd.DataFrame(dic_res, index=["test"]).astype(np.float32)
            df_runs.append(df_res)

            df_per_fold.append(pd.DataFrame(dic_res, index=[f"fold-{f:02d}"]).astype(np.float32))

            del dic_res
            gc.collect()

        aux_ = pd.concat(df_per_fold).reset_index()
        aux_["run"] = [f"run-{r:02d}" for _ in range(len(indexs_p_run_r))]
        df_per_run_fold.append(aux_.set_index(["run","index"]))

    #store per group
    save_results(f"{dir_folder}/plots/{ind_save}/results_all", pd.concat(df_per_run_fold))

    df_concat = pd.concat(df_runs).groupby(level=0)
    df_mean = df_concat.mean()
    df_std = df_concat.std()

    save_results(f"{dir_folder}/plots/{ind_save}/preds_mean", df_mean)
    save_results(f"{dir_folder}/plots/{ind_save}/preds_std", df_std)
    print(f"################ Showing the {ind_save} ################")
    print(df_mean.round(4).to_markdown())
    print(df_std.round(4).to_markdown())

    if task_type in ["classification", "multilabel"]:
        df_concat_diss = pd.concat(df_runs_diss).groupby(level=0)
        df_mean_diss = df_concat_diss.mean()
        df_std_diss = df_concat_diss.std()

        save_results(f"{dir_folder}/plots/{ind_save}/preds_ind_mean", df_mean_diss)
        save_results(f"{dir_folder}/plots/{ind_save}/preds_ind_std", df_std_diss)

    return df_mean,df_std

def calculate_metrics(df_summary, df_std, data_te,data_name, method, task_type="classification", **args):
    preds_p_run_te, indexs_p_run_te = load_data_per_fold(data_name, method, **args)

    df_aux, df_aux2= classification_metric(
                        preds_p_run_te,
                        indexs_p_run_te,
                        data_te,
                        ind_save=f"{data_name}/{method}/",
                        task_type = task_type,
                        **args
                        )
    df_summary[method] = df_aux.loc["test"]
    df_std[method] = df_aux2.loc["test"]

def main_evaluation(config_file):
    input_dir_folder = config_file["input_dir_folder"]
    output_dir_folder = config_file["output_dir_folder"]
    data_name = config_file["data_name"]

    if "train" in data_name:
        data_te = load_structure(input_dir_folder, data_name.replace("train", "test"), load_memory=config_file.get("load_memory", False))
    else:
        data_te = load_structure(input_dir_folder, data_name, load_memory=config_file.get("load_memory", False))

    if config_file.get("methods_to_plot"):
        methods_to_plot = config_file["methods_to_plot"]
    else:
        methods_to_plot = sorted(os.listdir(f"{output_dir_folder}/pred/{data_name}/"))

    df_summary_sup, df_summary_sup_s = pd.DataFrame(), pd.DataFrame()
    for method in methods_to_plot:
        print(f"Evaluating method {method}")
        calculate_metrics(df_summary_sup, df_summary_sup_s,
                        data_te,
                        data_name,
                        method,
                        dir_folder=output_dir_folder,
                        task_type = config_file.get("task_type", "classification"),
                        )
        gc.collect()

    #all figures were saved in output_dir_folder/plots
    print(">>>>>>>>>>>>>>>>> Mean across runs on test set")
    print((df_summary_sup.T).round(4).to_markdown())
    print(">>>>>>>>>>>>>>>>> Std across runs on test set")
    print((df_summary_sup_s.T).round(4).to_markdown())
    df_summary_sup.T.to_csv(f"{output_dir_folder}/plots/{data_name}/summary_mean.csv")
    df_summary_sup_s.T.to_csv(f"{output_dir_folder}/plots/{data_name}/summary_std.csv")

if __name__ == "__main__":
    arg_parser = argparse.ArgumentParser()
    arg_parser.add_argument(
        "--settings_file",
        "-s",
        action="store",
        dest="settings_file",
        required=True,
        type=str,
        help="path of the settings file",
    )
    args = arg_parser.parse_args()
    with open(args.settings_file) as fd:
        config_file = yaml.load(fd, Loader=yaml.SafeLoader)

    main_evaluation(config_file)