I have a dataframe like so.
id K V
0 1 k1 3
1 1 k2 4
2 1 k2 5
3 1 k1 5
4 2 k1 2
5 2 k1 3
6 2 k2 3
And i also have a set of conditions like k1 > 1 and k2 < 4.
I want to process the conditions and create a new dataframe containing 1 row per id and columns for each conditions.
id k1_condition k2_condition
0 1 True False
1 2 True True
Try the following
# function to be applied to the column 'V 'each (id, ki) group
def conditions(g):
cond_dict = {
'k1': lambda k1: k1 > 1,
'k2': lambda k2: k2 < 4
}
_ , k = g.name # g.name = group key = (id, ki)
return cond_dict[k](g).all()
out = (
df.groupby(['id', 'K'])['V']
.apply(conditions)
.unstack('K') # turn k1 and k2 into columns
.add_suffix('_cond') # add suffix to column names: ki --> ki_cond
.rename_axis(columns=None) # remove the column axis label (K)
.reset_index() # make id a column, not the index
)
Output:
>>> out
id k1_cond k2_cond
0 1 True False
1 2 True True
You can easily add more conditions to the cond_dict if the column K contains other values besides k1 and k2.
here is one way to do it
df2=df.pivot_table(index='id',columns='K', values='V', aggfunc=['min','max']).reset_index()
df2.columns = ['_'.join(col) for col in df2.columns ]
df2['k1_condition'] = df2['min_k1'] > 1
df2['k2_condition'] = df2['max_k2'] <4
df2=df2.drop(columns=['min_k1','min_k2','max_k1','max_k2'])
df2
OR
df2=df.pivot_table(index='id',columns='K', values='V', aggfunc=['min','max']).reset_index()
df2['k1_condition'] = df2['min']['k1'] > 1
df2['k2_condition'] = df2['max']['k2'] <4
df2.drop(columns=['min','max'],level=0,inplace=True)
df2
id_ k1_condition k2_condition
0 1 True False
1 2 True True
Dataframe.apply should work:
df["k1_condition"] = df.apply(lambda x: x["K"]=="k1" & x["V"]>1, axis=1)
df["k2_condition"] = df.apply(lambda x: x["K"]=="k2" & x["V"]>4, axis=1)
df2 = df[["id", "k1_condition", "k2_condition"]].groupy("id").any()
You could use pivot_table with a conditions function.
def conditions(x):
k = df.at[x.index[0],'K']
if k == 'k1':
return (x>1).all()
return (x<4).all()
pd.pivot_table(df, index='id', columns='K', aggfunc=conditions) \
.droplevel(level=0, axis=1).add_suffix('_condition') \
.rename_axis(None, axis=1).reset_index()
Result
id k1_conditon k2_condition
0 1 True False
1 2 True True
