Skip to content

Plotting

physXAI.plotting.plotting

Classes

Functions

subplots(header: str, *figures) -> go.Figure

Source code in physXAI/plotting/plotting.py 
 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
def subplots(header: str, *figures) -> go.Figure:
    subplot_titles = []
    plot_type = []

    # Define titles and specs
    for fig_dict in figures:
        subplot_titles.append(fig_dict["title"])

        plot_type.append([{"type": fig_dict["type"]}])

    # Create the subplots
    fig = make_subplots(rows=len(figures), cols=1, subplot_titles=subplot_titles, specs=plot_type)

    # Add each figure's traces to the subplots
    for i, fig_dict in enumerate(figures):
        figure = fig_dict["figure"]

        # For specifically table type figures
        for trace in figure["data"]:
            fig.add_trace(trace, row=i + 1, col=1)
        if fig_dict["type"] == "scatter":
            fig.update_xaxes(title_text=figure.layout.xaxis.title.text, row=i + 1, col=1)
            fig.update_yaxes(title_text=figure.layout.yaxis.title.text, row=i + 1, col=1)

    fig.update_layout(title_text=header, height=300*len(figures), width=1250)

    fig.show()

    return fig

plot_metrics_table(td: TrainingDataGeneric) -> go.Figure

Source code in physXAI/plotting/plotting.py 
40
41
42
43
44
45
46
47
48
49
50
51
52
def plot_metrics_table(td: TrainingDataGeneric) -> go.Figure:
    # Extracting metrics and corresponding values
    metrics = td.metrics
    labels, values = metrics.get_metrics()

    # Creating the table
    fig = go.Figure()
    fig.add_trace(go.Table(
            header=dict(values=['Metrics', 'Value']),
            cells=dict(values=[labels, values])
        ))

    return fig

plot_prediction_correlation(td: TrainingDataGeneric) -> go.Figure

Source code in physXAI/plotting/plotting.py 
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
def plot_prediction_correlation(td: TrainingDataGeneric) -> go.Figure:
    fig1 = go.Figure()

    x = np.linspace(np.min(td.y_train_single), np.max(td.y_train_single), 100)
    fig1.add_trace(go.Scatter(x=x, y=x, mode='lines', line=dict(color='black', dash='dash'), name='Ref'))

    fig1.add_trace(go.Scatter(x=td.y_train_single[:, 0], y=td.y_train_pred_single[:, 0], name='Train', mode='markers',
                              marker=dict(color='green')))
    if td.y_val_single is not None:
        fig1.add_trace(
            go.Scatter(x=td.y_val_single[:, 0], y=td.y_val_pred_single[:, 0], name='Val', mode='markers',
                       marker=dict(color='blue')))
    fig1.add_trace(go.Scatter(x=td.y_test_single[:, 0], y=td.y_test_pred_single[:, 0], name='Test', mode='markers',
                              marker=dict(color='red')))

    fig1.update_layout(
        xaxis_title="True Values",
        yaxis_title="Predicted Values"
    )
    return fig1

plot_predictions(td: TrainingDataGeneric) -> go.Figure

Source code in physXAI/plotting/plotting.py 
 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
def plot_predictions(td: TrainingDataGeneric) -> go.Figure:

    y_train = td.y_train_single[:, 0]
    y_train_pred = td.y_train_pred_single
    if td.y_val_single is not None:
        y_val = td.y_val_single[:, 0]
        y_val_pred = td.y_val_pred_single
    else:
        y_val = np.array([])
        y_val_pred = np.array([])
    y_test = td.y_test_single[:, 0]
    y_test_pred = td.y_test_pred_single

    # Plot the predictions using plotly library
    y_train_combined = list(zip(y_train, y_train_pred, ['Train'] * len(y_train)))
    y_val_combined = list(zip(y_val, y_val_pred, ['Val'] * len(y_val)))
    y_test_combined = list(zip(y_test, y_test_pred, ['Test'] * len(y_test)))
    combined = y_train_combined + y_val_combined + y_test_combined
    combined.sort(key=lambda x: x[0])
    y_sorted, y_pred_sorted, labels = zip(*combined)

    df_plotly = pd.DataFrame(combined, columns=['y_sorted', 'y_pred_sorted', 'labels'])
    df_plotly['y_pred_sorted'] = df_plotly['y_pred_sorted'].apply(
        lambda x: x[0] if isinstance(x, (list, np.ndarray)) else x)

    fig1 = go.Figure()
    fig1.add_trace(go.Scatter(x=list(range(len(y_sorted))), y=df_plotly['y_sorted'],
                              mode='markers',
                              name='True values',
                              marker=dict(color='black')))

    color_map = {
        'Train': 'green',
        'Val': 'blue',
        'Test': 'red',
    }

    fig2 = px.scatter(df_plotly, x=list(range(len(y_sorted))), y='y_pred_sorted', color='labels',
                      color_discrete_map=color_map)

    fig = go.Figure(data=fig1.data + fig2.data)

    fig.update_layout(
        xaxis_title="Sample",
        yaxis_title="Value"
    )

    return fig

plot_training_history(td: TrainingDataGeneric) -> go.Figure

Source code in physXAI/plotting/plotting.py 
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
def plot_training_history(td: TrainingDataGeneric) -> go.Figure:
    history = td.training_record.history
    # Plot training history using plotly library
    training_rmse = history['rmse']
    len_train = len(training_rmse)
    if td.y_val is not None:
        validation_rmse = history['val_rmse']
        len_val = len(validation_rmse)
    else:
        validation_rmse = []
        len_val = 0

    data = {
        'Epochs': list(range(1, len_train + 1)) + list(range(1, len_val + 1)),
        'RMSE': training_rmse + validation_rmse,
        'Type': ['Training RMSE'] * len_train + ['Validation RMSE'] * len_val
    }

    df = pd.DataFrame(data)

    color_map = {
        'Training RMSE': 'green',
        'Validation RMSE': 'blue',
    }

    fig = px.line(df, x='Epochs', y='RMSE', color='Type', markers=True, color_discrete_map=color_map)

    return fig

plot_multi_rmse(td: TrainingDataMultiStep) -> go.Figure

Source code in physXAI/plotting/plotting.py 
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
def plot_multi_rmse(td: TrainingDataMultiStep) -> go.Figure:
    fig1 = go.Figure()
    fig1.add_trace(go.Scatter(x=np.array(range(1, len(td.metrics.rmse_train_l)+1)), y=np.array(td.metrics.rmse_train_l),
                              name='Train', mode='lines', marker=dict(color='green')))
    if td.y_val is not None:
        fig1.add_trace(go.Scatter(x=np.array(range(1, len(td.metrics.rmse_val_l) + 1)),
                                  y=np.array(td.metrics.rmse_val_l), name='Val',
                                  mode='lines', marker=dict(color='blue')))
    fig1.add_trace(go.Scatter(x=np.array(range(1, len(td.metrics.rmse_test_l)+1)), y=np.array(td.metrics.rmse_test_l),
                              name='Test', mode='lines', marker=dict(color='red')))

    fig1.update_layout(
        xaxis_title="Prediction Step",
        yaxis_title="RMSE"
    )

    return fig1

plot_recFeatureSelection(fs: dict, multi_step: bool, use_multi_step_error: bool)

Source code in physXAI/plotting/plotting.py 
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
def plot_recFeatureSelection(fs: dict, multi_step: bool, use_multi_step_error: bool):

    x = list()
    y = list()
    y_single = list()

    for k, v in fs.items():

        x.append(k)
        values = list()
        values_single = list()
        for f in v:
            values.append(f['kpi'])
            if multi_step:
                values_single.append(f['kpi_single_step'])

        if not multi_step:
            y.append(min(values))
        elif use_multi_step_error:
            index = values.index(min(values))
            y.append(values[index])
            y_single.append(values_single[index])
        else:
            index = values_single.index(min(values_single))
            y.append(values[index])
            y_single.append(values_single[index])

    fig = go.Figure()
    if not multi_step:
        fig.add_trace(go.Scatter(x=x, y=y, mode='lines+markers', name='Single Step',
                                 marker=dict(color='blue')))
    else:
        fig.add_trace(go.Scatter(x=x, y=y, mode='lines+markers', name='Multi Step',
                                 marker=dict(color='blue')))
        fig.add_trace(go.Scatter(x=x, y=y_single, mode='lines+markers', name='Single Step',
                                 marker=dict(color='green')))

    fig.update_layout(
        xaxis_title="Number Features",
        yaxis_title="RMSE"
    )

    fig.show()