在材料信息学中Magpie特征获取

1.导入包

from matminer.featurizers.base import MultipleFeaturizer
from matminer.featurizers import composition as cf
from matminer.featurizers.conversions import StrToComposition
import pandas as pd

2.构建dataframe格式材料数据样本

   用dataframe储存材料数据，至少要有材料的化学式，用于得到材料的Magpie特征；

  这里提供一个函数得到材料的composition；

def get_magpie_composition(formula):
    upper_letter = ['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 'S', 'T',
                    'U', 'V', 'W', 'X', 'Y', 'Z']
    number = ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9']

    ins = []
    for i in range(len(formula)):
        if i == 0:
            continue
        if formula[i] in upper_letter:
            if formula[i - 1] not in number:
                ins.append(i)

    formula_list = list(formula)
    k = 0
    for j in ins:
        formula_list.insert(j + k, '1')
        k = k + 1

    formula = ''.join(formula_list)

    if formula[len(formula) - 1] not in number:
        formula = formula + '1'
    return formula

  对dataframe中的每个材料化学式执行get_magpie_composition()函数，并且将执行结果储存到dataframe中的列composition中，得到如下dataframe。

3. 计算Magpie特征

"""
需要输入材料的化学式(例如Nb1Pt1Si1)，可以得到材料的145维向量表示
"""

def magpie(df,composition):
    df = StrToComposition(target_col_id='composition_obj').featurize_dataframe(df, composition)
    feature_calculators = MultipleFeaturizer([cf.Stoichiometry(), cf.ElementProperty.from_preset("magpie"),
                                              cf.ValenceOrbital(props=['avg']), cf.IonProperty(fast=True)])

    feature_labels = feature_calculators.feature_labels()
    print('feature_labels')
    print(feature_labels)

    df = feature_calculators.featurize_dataframe(df, col_id='composition_obj')

    return df

  对dataframe利用上述函数可以得到材料的145维特征，储存在原dataframe中。