python - 将 Pandas 数据框转换为橙色数据表

Question

我正在处理 pandas 数据框，我想将其转换为橙色数据表以估算缺失值。我的数据框看起来像

locationId rank Rating type  value
1          1    10     shop   2.668
2          4    8      store  3.921
3          3    NAN    shop   3.122

其中 rank 是有序的，重复值在 1 到 5 之间。Type 是分类的，类型为 shop、store 等，rating 是整数类型。value 是 float 。我想将它转换成橙色数据表并估算缺失值。我也看过this但是这些函数给我一个错误，对我不起作用。

Answer 1

import numpy as np
import pandas as pd
import Orange
import csv
from io import StringIO
from collections import OrderedDict
from Orange.data import Table, Domain, ContinuousVariable, DiscreteVariable


def pandas_to_orange(df):
    domain, attributes, metas = construct_domain(df)
    orange_table = Orange.data.Table.from_numpy(domain = domain, X = df[attributes].values, Y = None, metas = df[metas].values, W = None)
    return orange_table

def construct_domain(df):
    columns = OrderedDict(df.dtypes)
    attributes = OrderedDict()
    metas = OrderedDict()
    for name, dtype in columns.items():

        if issubclass(dtype.type, np.number):
            if len(df[name].unique()) >= 13 or issubclass(dtype.type, np.inexact) or (df[name].max() > len(df[name].unique())):
                attributes[name] = Orange.data.ContinuousVariable(name)
            else:
                df[name] = df[name].astype(str)
                attributes[name] = Orange.data.DiscreteVariable(name, values = sorted(df[name].unique().tolist()))
        else:
            metas[name] = Orange.data.StringVariable(name)

    domain = Orange.data.Domain(attributes = attributes.values(), metas = metas.values())

    return domain, list(attributes.keys()), list(metas.keys())

使用 -> pandas_to_orange(dataFrame)