File size: 3,026 Bytes
a3c36fd
 
4207871
a3c36fd
b7541ec
 
d98720a
4207871
e57f927
bc45490
 
 
 
 
 
 
b7541ec
 
 
 
 
 
 
bc45490
b7541ec
 
 
 
 
 
 
 
 
 
 
 
 
 
bc45490
a3c36fd
 
aa3260f
a3c36fd
 
 
 
 
 
bc45490
 
 
e57f927
bc45490
 
 
 
 
 
 
d98720a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
bc45490
b7541ec
 
 
 
e57f927
 
bc45490
 
 
3cd0d7b
bc45490
7e5c64b
 
 
bc45490
 
33354da
bc45490
 
7e5c64b
bc45490
 
33354da
bc45490
7e5c64b
bc45490
 
33354da
bc45490
 
7e5c64b
33354da
 
7e5c64b
bc45490
 
4207871
bc45490
4207871
a3c36fd
bc45490
 
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
import io
import gradio as gr
import os
import tempfile
import numpy as np
import pandas as pd
import traceback as tb


def greet(
    file_obj: tempfile._TemporaryFileWrapper,
    col_to_fit: str,
    niterations: int,
    binary_operators: list,
    unary_operators: list,
):
    empty_df = pd.DataFrame(
        {
            "equation": [],
            "loss": [],
            "complexity": [],
        }
    )
    if col_to_fit == "":
        return (
            empty_df,
            "Please enter a column to predict!",
        )
    if len(binary_operators) == 0 and len(unary_operators) == 0:
        return (
            empty_df,
            "Please select at least one operator!",
        )
    if file_obj is None:
        return (
            empty_df,
            "Please upload a CSV file!",
        )
    niterations = int(niterations)
    # Need to install PySR in separate python instance:
    os.system(
        """if [ ! -d "$HOME/.julia/environments/pysr-0.9.3" ]
    then
        python -c 'import pysr; pysr.install()'
    fi"""
    )
    from pysr import PySRRegressor

    df = pd.read_csv(file_obj.name)
    y = np.array(df[col_to_fit])
    X = df.drop([col_to_fit], axis=1)

    model = PySRRegressor(
        update=False,
        temp_equation_file=True,
        niterations=niterations,
        binary_operators=binary_operators,
        unary_operators=unary_operators,
    )
    try:
        model.fit(X, y)
    # Catch all error:
    except Exception as e:
        error_traceback = tb.format_exc()
        if "CalledProcessError" in error_traceback:
            return (
                empty_df,
                "Could not initialize Julia. Error message:\n"
                + error_traceback,
            )
        else:
            return (
                empty_df,
                "Failed due to error:\n" + error_traceback,
            )

    df = model.equations_[["equation", "loss", "complexity"]]
    # Convert all columns to string type:
    df = df.astype(str)
    return df, "Successful."


def main():
    demo = gr.Interface(
        fn=greet,
        description="PySR Demo",
        inputs=[
            gr.inputs.File(label="Upload a CSV File"),
            gr.inputs.Textbox(label="Column to Predict", placeholder="y"),
            gr.inputs.Slider(
                minimum=1,
                maximum=1000,
                default=40,
                label="Number of iterations",
            ),
            gr.inputs.CheckboxGroup(
                choices=["+", "-", "*", "/", "^"],
                label="Binary Operators",
                default=["+", "-", "*", "/"],
            ),
            gr.inputs.CheckboxGroup(
                choices=["sin", "cos", "exp", "log"],
                label="Unary Operators",
                default=[],
            ),
        ],
        outputs=[
            "dataframe",
            gr.outputs.Textbox(label="Error Log"),
        ],
    )
    # Add file to the demo:

    demo.launch()


if __name__ == "__main__":
    main()