I am struggling to replicate sas(another programming language) inner merge in python .
The python inner merge is not matching with sas inner merge when duplicate key values are coming .
Below is an example :
zw = pd.DataFrame({"ID":[1,0,0,1,0,0,1],
"Name":['Shivansh','Shivansh','Shivansh','Amar','Arpit','Ranjeet','Priyanka'],
"job_profile":['DataS','SWD','DataA','DataA','AndroidD','PythonD','fullstac'],
"salary":[22,15,10,9,16,18,22],
"city":['noida','bangalore','hyderabad','noida','pune','gurugram','bangalore'],
"ant":[10,15,15,10,16,17,18]})
zw1 = pd.DataFrame({"ID-":[1,0,0,1,0,0,1],
"Name":['Shivansh','Shivansh','Swati','Amar','Arpit','Ranjeet','Priyanka'],
"job_profile_":['DataS','SWD','DataA','DataA','AndroidD','PythonD','fullstac'],
"salary_":[2,15,10,9,16,18,22],
"city_":['noida','kochi','hyderabad','noida','pune','gurugram','bangalore'],
"ant_":[1,15,15,10,16,17,18]})
zw and sw1 are the input tables . Both the tables need to be inner merged on the key column Name .The issue is both columns are having duplicate values in Name column .
Python is generating all possible combinations with the duplicate rows .
Below is the expected output :
I tried normal inner merge and tried dropping duplicate row with ID and Name columns , but still not getting the desired output .
df1=pd.merge(zw,zw1,on=['Name'],how='inner')
df1.drop_duplicates(['Name','ID'])
Use df.combine_first + df.sort_values combination:
df = zw.combine_first(zw1).sort_values('Name')
print(df)
ID ID- Name ant ant_ city city_ job_profile \
3 1 1 Amar 10 10 noida noida DataA
4 0 0 Arpit 16 16 pune pune AndroidD
6 1 1 Priyanka 18 18 bangalore bangalore fullstac
5 0 0 Ranjeet 17 17 gurugram gurugram PythonD
0 1 1 Shivansh 10 1 noida noida DataS
1 0 0 Shivansh 15 15 bangalore kochi SWD
2 0 0 Shivansh 15 15 hyderabad hyderabad DataA
job_profile_ salary salary_
3 DataA 9 9
4 AndroidD 16 16
6 fullstac 22 22
5 PythonD 18 18
0 DataS 22 2
1 SWD 15 15
2 DataA 10 10
Related
I have two datasets, each about half a million observations. I am writing the below code and it seems the code never seems to stop executing. I would like to know if there is a better way of doing it. Appreciate inputs.
Below are sample formats of my dataframes. Both dataframes share a set of 'sid' values , meaning all the 'sid' values in 'df2' will have a match in 'df1' 'sid' values. The 'tid' values and consequently the 'rid' values (which are a combination of 'sid' and 'tid' values) may not appear in both sets.
The task is simple. I would like to create the 'tv' column in df2. Wherever the 'rid' in df2 matches with the 'rid' in 'df1', the 'tv' column in df2 takes the corresponding 'tv' value from df1. If it does not match, the 'tv' value in 'df2' will be the median 'tv' value for the matching 'sid' subset in 'df1'.
In fact my original task includes creating a few more similar columns like 'tv' in df2 (based on their values in 'df1' ; these columns exist in 'df1').
I believe as my code contains for loop combined with if else statement and multiple value assignment statements, it is taking forever to execute. Appreciate any inputs.
df1
sid tid rid tv
0 0 0 0-0 9
1 0 1 0-1 8
2 0 3 0-3 4
3 1 5 1-5 2
4 1 7 1-7 3
5 1 9 1-9 14
6 1 10 1-10 24
7 1 11 1-11 13
8 2 14 2-14 2
9 2 16 2-16 5
10 3 17 3-17 6
11 3 18 3-18 8
12 3 20 3-20 5
13 3 21 3-21 11
14 4 23 4-23 6
df2
sid tid rid
0 0 0 0-0
1 0 2 0-2
2 1 3 1-3
3 1 6 1-6
4 1 9 1-9
5 2 10 2-10
6 2 12 2-12
7 3 1 3-1
8 3 15 3-15
9 3 1 3-1
10 4 19 4-19
11 4 22 4-22
rids = [rid.split('-') for rid in df1.rid]
for r in df2.rid:
s,t = r.split('-')
if [s,t] in rids:
df2.loc[df2.rid== r,'tv'] = df1.loc[df1.rid == r,'tv']
else:
df2.loc[df2.rid== r,'tv'] = df1.loc[df1.sid == int(s),'tv'].median()
The expected df2 shall be as follows:
sid tid rid tv
0 0 0 0-0 9.0
1 0 2 0-2 8.0
2 1 3 1-3 13.0
3 1 6 1-6 13.0
4 1 9 1-9 14.0
5 2 10 2-10 3.5
6 2 12 2-12 3.5
7 3 1 3-1 7.0
8 3 15 3-15 7.0
9 3 1 3-1 7.0
10 4 19 4-19 6.0
11 4 22 4-22 6.0
You can left merge on df2 with a subset(because you need only tv column you can also pass the df1 without any subset) of df1 on 'rid' then calculate median and fill values:
out=df2.merge(df1[['rid','tv']],on='rid',how='left')
out['tv']=out['tv_y'].fillna(out['sid'].map(df1.groupby('sid')['tv'].median()))
out= out.drop(['tv_x','tid_y','tv_y'], axis=1)
out = out.rename(columns = {'tid_x': 'tid'})
out
OR
Since you said that:
all the 'sid' values in 'df2' will have a match in 'df1' 'sid' values
So you can also left merge them on ['sid','rid'] and then fillna() value of tv with the median of df1 'tv' column by mapping values using map() method:
out=df2.merge(df1,on=['sid','rid'],how='left')
out['tv']=out['tv_y'].fillna(out['sid'].map(df1.groupby('sid')['tv'].median()))
out= out.drop(['tv_x','tv_y'], axis=1)
out
output of out:
sid tid rid tv
0 0 0 0-0 9.0
1 0 2 0-2 8.0
2 1 3 1-3 13.0
3 1 6 1-6 13.0
4 1 9 1-9 14.0
5 2 10 2-10 3.5
6 2 12 2-12 3.5
7 3 1 3-1 7.0
8 3 15 3-15 7.0
9 3 1 3-1 7.0
10 4 19 4-19 6.0
11 4 22 4-22 6.0
Here is a suggestion without any loops, based on dictionaries:
matching_values = dict(zip(df1['rid'][df1['rid'].isin(df2['rid'])], df1['tv'][df1['rid'].isin(df2['rid'])]))
df2[df2['rid'].isin(df1['rid'])]['tv'] = df2[df2['rid'].isin(df1['rid'])]['rid']
df2[df2['rid'].isin(df1['rid'])]['tv'].replace(matching_values)
median_values = df2[(~df2['rid'].isin(df1['rid']) & (df2['sid'].isin(df1['sid'])].groupby('sid')['tv'].median().to_dict()
df2[(~df2['rid'].isin(df1['rid']) & (df2['sid'].isin(df1['sid'])]['tv'] = df2[(~df2['rid'].isin(df1['rid']) & (df2['sid'].isin(df1['sid'])]['sid']
df2[(~df2['rid'].isin(df1['rid']) & (df2['sid'].isin(df1['sid'])]['tv'].replace(median_values)
This should do the trick. The logic here is that we first create a dictionary, in which the "rid and "sid" values are the keys and the median and matching "tv" values are the dictionary values. Next, we replace the "tv" values in df2 with the rid and sid keys, respectively, (because they are the dictionary keys) which can thus easily be replaced by the correct tv values by calling .replace().
Don't use for loops in pandas, that is known to be slow. That way you don't get to benefit from all the internal optimizations that have been made.
Try to use the split-apply-combine pattern:
split df1 into sid to calculate the median: df1.groupby('sid')['tv'].median()
join df2 on df1: df2.join(df1.set_index('rid'), on='rid')
fill the NaN values with the median calculated in step 1.
(Haven't tested the code).
Let's say I have students infos id, age and class as follows:
id age class
0 1 23 a
1 2 24 a
2 3 25 b
3 4 22 b
4 5 16 c
5 6 16 d
I want to groupby class and create a new column named major by randomly assign math, art, business, science into it, which means for same class, the major strings are same.
We may need to use apply(lambda x: random.choice..) to realize this, but I don't know how to do this. Thanks for your help.
Output expected:
id age major class
0 1 23 art a
1 2 24 art a
2 3 25 science b
3 4 22 science b
4 5 16 business c
5 6 16 math d
Use numpy.random.choice with number of values by length of DataFrame:
df['major'] = np.random.choice(['math', 'art', 'business', 'science'], size=len(df))
print (df)
id age major
0 1 23 business
1 2 24 art
2 3 25 science
3 4 22 math
4 5 16 science
5 6 16 business
EDIT: for same major values per groups use Series.map with dictionary:
c = df['class'].unique()
vals = np.random.choice(['math', 'art', 'business', 'science'], size=len(c))
df['major'] = df['class'].map(dict(zip(c, vals)))
print (df)
id age class major
0 1 23 a business
1 2 24 a business
2 3 25 b art
3 4 22 b art
4 5 16 c science
5 6 16 d math
in my dataframe have three columns columns value ,ID and distance . i want to check in ID column when its changes from 2 to any other value count rows and record first value and last value when 2 changes to other value and save and also save corresponding value of column distance when change from 2 to other in ID column.
df=pd.DataFrame({'value':[3,4,7,8,11,20,15,20,15,16],'ID':[2,2,8,8,8,2,2,2,5,5],'distance':[0,0,1,0,0,0,0,0,0,0]})
print(df)
value ID distance
0 3 2 0
1 4 2 0
2 7 8 1
3 8 8 0
4 11 8 0
5 20 2 0
6 15 2 0
7 20 2 0
8 15 5 0
9 16 5 0
required results:
df_out=pd.DataFrame({'rows_Count':[3,2],'value_first':[7,15],'value_last':[11,16],'distance_first':[1,0]})
print(df_out)
rows_Count value_first value_last distance_first
0 3 7 11 1
1 2 15 16 0
Use:
#compare by 2
m = df['ID'].eq(2)
#filter out data before first 2 (in sample data not, in real data possible)
df = df[m.cumsum().ne(0)]
#create unique groups for non 2 groups, add misisng values by reindex
s = m.ne(m.shift()).cumsum()[~m].reindex(df.index)
#aggregate with helper s Series
df1 = df.groupby(s).agg({'ID':'size', 'value':['first','last'], 'distance':'first'})
#flatten MultiIndex
df1.columns = df1.columns.map('_'.join)
df1 = df1.reset_index(drop=True)
print (df1)
ID_size value_first value_last distance_first
0 3 7 11 1
1 2 15 16 0
Verify in changed data (not only 2 first group):
df=pd.DataFrame({'value':[3,4,7,8,11,20,15,20,15,16],
'ID':[1,7,8,8,8,2,2,2,5,5],
'distance':[0,0,1,0,0,0,0,0,0,0]})
print(df)
value ID distance
0 3 1 0 <- changed ID
1 4 7 0 <- changed ID
2 7 8 1
3 8 8 0
4 11 8 0
5 20 2 0
6 15 2 0
7 20 2 0
8 15 5 0
9 16 5 0
#compare by 2
m = df['ID'].eq(2)
#filter out data before first 2 (in sample data not, in real data possible)
df = df[m.cumsum().ne(0)]
#create unique groups for non 2 groups, add misisng values by reindex
s = m.ne(m.shift()).cumsum()[~m].reindex(df.index)
#aggregate with helper s Series
df1 = df.groupby(s).agg({'ID':'size', 'value':['first','last'], 'distance':'first'})
#flatten MultiIndex
df1.columns = df1.columns.map('_'.join)
df1 = df1.reset_index(drop=True)
print (df1)
ID_size value_first value_last distance_first
0 2 15 16 0
I have an dataframe which contain the observed data as:
import pandas as pd
d = {'humanID': [1, 1, 2,2,2,2 ,2,2,2,2], 'dogID':
[1,2,1,5,4,6,7,20,9,7],'month': [1,1,2,3,1,2,3,1,2,2]}
df = pd.DataFrame(data=d)
The df is follow
humanID dogID month
0 1 1 1
1 1 2 1
2 2 1 2
3 2 5 3
4 2 4 1
5 2 6 2
6 2 7 3
7 2 20 1
8 2 9 2
9 2 7 2
We total have two human and twenty dog, and above df contains the observed data. For example:
The first row means: human1 adopt dog1 at January
The second row means: human1 adopt dog2 at January
The third row means: human2 adopt dog1 at Febuary
========================================================================
My goal is randomly generating two unobserved data for each (human, month) that are not appear in the original observed data.
like for human1 at January, he does't adopt the dog [3,4,5,6,7,..20] And I want to randomly create two unobserved sample (human, month) in triple form
humanID dogID month
1 20 1
1 10 1
However, the follow sample is not allowed since it appear in original df
humanID dogID month
1 2 1
For human1, he doesn't have any activity at Feb, so we don't need to sample the unobserved data.
For human2, he have activity for Jan, Feb and March. Therefore, for each month, we want to randomly create the unobserved data. For example, In Jan, human2 adopt dog1, dog4 and god 20. The two random unobserved samples can be
humanID dogID month
2 2 1
2 6 1
same process can be used for Feb and March.
I want to put all of the unobserved in one dataframe such as follow unobserved
humanID dogID month
0 1 20 1
1 1 10 1
2 2 2 1
3 2 6 1
4 2 13 2
5 2 16 2
6 2 1 3
7 2 20 3
Any fast way to do this?
PS: this is a code interview for a start-up company.
Using groupby and random.choices:
import random
dogs = list(range(1,21))
dfs = []
n_sample = 2
for i,d in df.groupby(['humanID', 'month']):
h_id, month = i
sample = pd.DataFrame([(h_id, dogID, month) for dogID in random.choices(list(set(dogs)-set(d['dogID'])), k=n_sample)])
dfs.append(sample)
new_df = pd.concat(dfs).reset_index(drop=True)
new_df.columns = ['humanID', 'dogID', 'month']
print(new_df)
humanID dogID month
0 1 11 1
1 1 5 1
2 2 19 1
3 2 18 1
4 2 15 2
5 2 14 2
6 2 16 3
7 2 18 3
If I understand you correctly, you can use np.random.permutation() for the dogID column to generate random permutations of the column,
df_new=df.copy()
df_new['dogID']=np.random.permutation(df.dogID)
print(df_new.sort_values('month'))
humanID dogID month
0 1 1 1
1 1 20 1
4 2 9 1
7 2 1 1
2 2 4 2
5 2 5 2
8 2 2 2
9 2 7 2
3 2 7 3
6 2 6 3
Or to create random sampling of missing values within the range of dogID:
df_new=df.copy()
a=np.random.permutation(range(df_new.dogID.min(),df_new.dogID.max()))
df_new['dogID']=np.random.choice(a,df_new.shape[0])
print(df_new.sort_values('month'))
humanID dogID month
0 1 18 1
1 1 16 1
4 2 1 1
7 2 8 1
2 2 4 2
5 2 2 2
8 2 16 2
9 2 14 2
3 2 4 3
6 2 12 3
With the DataFrame below as an example,
In [83]:
df = pd.DataFrame({'A':[1,1,2,2],'B':[1,2,1,2],'values':np.arange(10,30,5)})
df
Out[83]:
A B values
0 1 1 10
1 1 2 15
2 2 1 20
3 2 2 25
What would be a simple way to generate a new column containing some aggregation of the data over one of the columns?
For example, if I sum values over items in A
In [84]:
df.groupby('A').sum()['values']
Out[84]:
A
1 25
2 45
Name: values
How can I get
A B values sum_values_A
0 1 1 10 25
1 1 2 15 25
2 2 1 20 45
3 2 2 25 45
In [20]: df = pd.DataFrame({'A':[1,1,2,2],'B':[1,2,1,2],'values':np.arange(10,30,5)})
In [21]: df
Out[21]:
A B values
0 1 1 10
1 1 2 15
2 2 1 20
3 2 2 25
In [22]: df['sum_values_A'] = df.groupby('A')['values'].transform(np.sum)
In [23]: df
Out[23]:
A B values sum_values_A
0 1 1 10 25
1 1 2 15 25
2 2 1 20 45
3 2 2 25 45
I found a way using join:
In [101]:
aggregated = df.groupby('A').sum()['values']
aggregated.name = 'sum_values_A'
df.join(aggregated,on='A')
Out[101]:
A B values sum_values_A
0 1 1 10 25
1 1 2 15 25
2 2 1 20 45
3 2 2 25 45
Anyone has a simpler way to do it?
This is not so direct but I found it very intuitive (the use of map to create new columns from another column) and can be applied to many other cases:
gb = df.groupby('A').sum()['values']
def getvalue(x):
return gb[x]
df['sum'] = df['A'].map(getvalue)
df
In [15]: def sum_col(df, col, new_col):
....: df[new_col] = df[col].sum()
....: return df
In [16]: df.groupby("A").apply(sum_col, 'values', 'sum_values_A')
Out[16]:
A B values sum_values_A
0 1 1 10 25
1 1 2 15 25
2 2 1 20 45
3 2 2 25 45