def generate_data(name):
#这个函数作用，输入是指定的文件名，输出增加了gaps，is_daoti，以及其他共计145特征的完整向量矩阵
#name='test_plus_gaps.csv'
df=pd.read_csv(name,index_col=[0])
df['gaps']=-10.0
df_gap=pd.read_csv("gaps.csv",index_col = [0])
print(df_gap.index)
i=0
str_s=""
for j in range(len(df_gap.index)):
#先打印二者的id
# print(df.index[i])
str_s='mp-'+str(df_gap.index[j])
if(str_s==df.index[i]):
df.iloc[i,-1]=df_gap.iloc[j,0]
i=i+1
#print("确实一样")
print("合并完毕")
#同样的方法我们来建立不同的分类
df['is_daoti']=-2
for i in range(len(df.index)):
if(df.ix[i,-2]==0):
df.ix[i,-1]=1
else:
df.ix[i,-1]=0
print("分类feature建立完成")
#首先使用describe获得对于数据的整体把握
print(df.describe())
df.describe().to_csv('general_look_jie.csv')
#通过观察数据发现并没有什么异常之处
df=StrToComposition().featurize_dataframe(df,'full_formula',ignore_errors=True)
print(df.head())
#print(df['composition'])
ep_feat=ElementProperty.from_preset(preset_name='magpie')
df=ep_feat.featurize_dataframe(df,col_id='composition',ignore_errors=True)#将composition这一列作为特征化的输入
print(df.head())
#print(ep_feat.citations())
#df.to_csv("plus the composition.csv")
#以上这部分是将formula转化为composition并转化feature
df=CompositionToOxidComposition().featurize_dataframe(df,col_id='composition')#引入了氧化态的相关特征
os_feat=OxidationStates()
df=os_feat.featurize_dataframe(df,col_id='composition_oxid')
new_name='2d_vector_plus.csv'
df.to_csv(new_name)
def test_featurizers():
df = pd.read_csv('test.csv', index_col=[0])
df = StrToComposition().featurize_dataframe(df, 'formula')
print(df.head())
#下一步，我们需要其中一个特征化来增加一系列的特征算符
ep_feat = ElementProperty.from_preset(preset_name='magpie')
df = ep_feat.featurize_dataframe(
df, col_id='composition') #将composition这一列作为特征化的输入
print(df.head())
print(ep_feat.citations())
#df.to_csv('将composition特征化后.csv')
#开始引入新的特征化算符吧
df = CompositionToOxidComposition().featurize_dataframe(
df, 'composition') #引入了氧化态的相关特征
os_feat = OxidationStates()
df = os_feat.featurize_dataframe(df, col_id='composition_oxid')
print(df.head())
df.to_csv('after_test.csv')
def generate_data():
df = load_elastic_tensor()
df.to_csv('原始elastic数据.csv')
print(df.columns)
unwanted_columns = [
'volume', 'nsites', 'compliance_tensor', 'elastic_tensor',
'elastic_tensor_original', 'K_Voigt', 'G_Voigt', 'K_Reuss', 'G_Reuss'
]
df = df.drop(unwanted_columns, axis=1)
print(df.head())
df.to_csv('扔掉不需要的部分.csv')
#首先使用describe获得对于数据的整体把握
print(df.describe())
df.describe().to_csv('general_look.csv')
#通过观察数据发现并没有什么异常之处
df = StrToComposition().featurize_dataframe(df, 'formula')
print(df.head())
df.to_csv('引入composition.csv')
#下一步，我们需要其中一个特征化来增加一系列的特征算符
ep_feat = ElementProperty.from_preset(preset_name='magpie')
df = ep_feat.featurize_dataframe(
df, col_id='composition') #将composition这一列作为特征化的输入
print(df.head())
print(ep_feat.citations())
df.to_csv('将composition特征化后.csv')
#开始引入新的特征化算符吧
df = CompositionToOxidComposition().featurize_dataframe(
df, 'composition') #引入了氧化态的相关特征
os_feat = OxidationStates()
df = os_feat.featurize_dataframe(df, col_id='composition_oxid')
print(df.head())
df.to_csv('引入氧化态之后.csv')
#其实除了基于composition的特征之外还有很多其他的，比如基于结构的
df_feat = DensityFeatures()
df = df_feat.featurize_dataframe(df, 'structure')
print(df.head())
df.to_csv('引入结构中的密度.csv')
print(df_feat.feature_labels())
data = [['mp-1025496', 'Nb1 Se2'], ['mp-977563', 'Nb1 Ir2'],
['mp-864631', 'Nb1 Rh2'], ['mp-3368', 'Nb3 O8']]
fdf = pd.DataFrame(data, columns=['Id', 'Reduced Formula'])
## Initial conversion to matminer objects
from matminer.featurizers.conversions import StrToComposition
fdf = StrToComposition().featurize_dataframe(fdf, 'Reduced Formula')
from matminer.featurizers.conversions import CompositionToOxidComposition
fdf = CompositionToOxidComposition().featurize_dataframe(fdf, 'composition')
print("The initial dataset has {}".format(fdf.shape))
print(fdf.head())
'''
Block 2 - Featurization
'''
#
# -- start F1
from matminer.featurizers.composition import ElementProperty
ep_feat = ElementProperty.from_preset(preset_name='magpie')
fdf = ep_feat.featurize_dataframe(fdf,
col_id='composition',
ignore_errors=True)
# Excluding the 'MagpieData' string from the columns' names
magpie_cols = [col for col in fdf.columns if 'MagpieData' in col]
if quick_demo:
query_string += " AND mdf.scroll_id:<10000"
data = mdf.get_data(query_string, unwind_arrays=False)
print(data.head())
# 重命名、预处理和筛选,delta_e应该是形成能
data = data[['oqmd.delta_e.value', 'material.composition']]
data = data.rename(columns={
'oqmd.delta_e.value': 'delta_e',
'material.composition': 'composition'
})
data = StrToComposition(
target_col_id='composition_obj').featurize_dataframe(
data, 'composition')
data.sort_values('delta_e', ascending=True, inplace=True)
print(data.head(3))
for k in ['delta_e']:
data[k] = pd.to_numeric(data[k])
original_count = len(data)
data = data[~data['delta_e'].isnull()]
print('Removed %d/%d entries' %
(original_count - len(data), original_count))
original_count = len(data)
data['composition'] = data['composition_obj'].apply(
lambda x: x.reduced_formula)
data.sort_values('delta_e', ascending=True, inplace=True)
data.drop_duplicates('composition', keep='first', inplace=True)
print('Removed %d/%d entries' %
(original_count - len(data), original_count))
