Function to determine window in a rolling function - python

I have a dataframe in which I want to apply a rolling mean over a column of numbers that come in 3-pairs where I only want 4 unique values to go into the mean.
Lets say my dataframe looks like:
Group Column to roll
1 9
2 5
2 5
2 4
2 4
2 4
2 3
2 3
2 3
2 6
2 6
2 6
2 8
Since I want 4 unique values to go into the mean but all values to be of equal weight and within the same group, my expected output (assuming I need 4 unique values) would be:
Group Output
1 nan
2 nan
2 nan
2 nan
2 nan
2 nan
2 nan
2 nan
2 nan
2 (6+3+4+5)/4
2 (6+3+4+5)/4
2 (6+3+4+5)/4
2 (8+6+3+4)/4
Any ideas how to do this?


You could try something like this:
df['Column to roll'].drop_duplicates().rolling(4).mean().reindex(df.index).ffill()
Output:
0 NaN
1 NaN
2 NaN
3 NaN
4 NaN
5 NaN
6 NaN
7 NaN
8 4.50
9 4.50
10 4.50
11 5.25
Name: Column to roll, dtype: float64
Edit question changed
df_out = df.groupby('Group')['Column to roll']\
.apply(lambda x: x.drop_duplicates().rolling(4).mean()).rename('Output')
df.set_index('Group',append=True).swaplevel(0,1)\
.join(df_out, how='left').ffill().reset_index(level=1, drop=True)
Output:
Column to roll Output
Group
1 9 NaN
2 5 NaN
2 5 NaN
2 4 NaN
2 4 NaN
2 4 NaN
2 3 NaN
2 3 NaN
2 3 NaN
2 6 4.50
2 6 4.50
2 6 4.50
2 8 5.25

Related

Concatenate columns skipping pasted rows and columns

I expect to describe well want I need. I have a data frame with the same columns name and another column that works as an index. The data frame looks as follows:
df = pd.DataFrame({'ID':[1,1,1,1,1,2,2,2,3,3,3,3],'X':[1,2,3,4,5,2,3,4,1,3,4,5],'Y':[1,2,3,4,5,2,3,4,5,4,3,2]})
df
Out[21]:
ID X Y
0 1 1 1
1 1 2 2
2 1 3 3
3 1 4 4
4 1 5 5
5 2 2 2
6 2 3 3
7 2 4 4
8 3 1 5
9 3 3 4
10 3 4 3
11 3 5 2
My intention is to copy X as an index or one column (it doesn't matter) and append Y columns from each 'ID' in the following way:

You can try
out = pd.concat([group.rename(columns={'Y': f'Y{name}'}) for name, group in df.groupby('ID')])
out.columns = out.columns.str.replace(r'\d+$', '', regex=True)
print(out)
ID X Y Y Y
0 1 1 1.0 NaN NaN
1 1 2 2.0 NaN NaN
2 1 3 3.0 NaN NaN
3 1 4 4.0 NaN NaN
4 1 5 5.0 NaN NaN
5 2 2 NaN 2.0 NaN
6 2 3 NaN 3.0 NaN
7 2 4 NaN 4.0 NaN
8 3 1 NaN NaN 5.0
9 3 3 NaN NaN 4.0
10 3 4 NaN NaN 3.0
11 3 5 NaN NaN 2.0

Here's another way to do it:
df_org = pd.DataFrame({'ID':[1,1,1,1,1,2,2,2,3,3,3,3],'X':[1,2,3,4,5,2,3,4,1,3,4,5]})
df = df_org.copy()
for i in set(df_org['ID']):
df1 = df_org[df_org['ID']==i]
col = 'Y'+str(i)
df1.columns = ['ID', col]
df = pd.concat([ df, df1[[col]] ], axis=1)
df.columns = df.columns.str.replace(r'\d+$', '', regex=True)
print(df)
Output:
ID X Y Y Y
0 1 1 1.0 NaN NaN
1 1 2 2.0 NaN NaN
2 1 3 3.0 NaN NaN
3 1 4 4.0 NaN NaN
4 1 5 5.0 NaN NaN
5 2 2 NaN 2.0 NaN
6 2 3 NaN 3.0 NaN
7 2 4 NaN 4.0 NaN
8 3 1 NaN NaN 1.0
9 3 3 NaN NaN 3.0
10 3 4 NaN NaN 4.0
11 3 5 NaN NaN 5.0

Another solution could be as follow.
Get unique values for column ID (stored in array s).
Use np.transpose to repeat column ID n times (n == len(s)) and evaluate the array's matches with s.
Use np.where to replace True with values from df.Y and False with NaN.
Finally, drop the orignal df.Y and rename the new columns as required.
import pandas as pd
import numpy as np
df = pd.DataFrame({'ID':[1,1,1,1,1,2,2,2,3,3,3,3],
'X':[1,2,3,4,5,2,3,4,1,3,4,5],
'Y':[1,2,3,4,5,2,3,4,5,4,3,2]})
s = df.ID.unique()
df[s] = np.where((np.transpose([df.ID]*len(s))==s),
np.transpose([df.Y]*len(s)),
np.nan)
df.drop('Y', axis=1, inplace=True)
df.rename(columns={k:'Y' for k in s}, inplace=True)
print(df)
ID X Y Y Y
0 1 1 1.0 NaN NaN
1 1 2 2.0 NaN NaN
2 1 3 3.0 NaN NaN
3 1 4 4.0 NaN NaN
4 1 5 5.0 NaN NaN
5 2 2 NaN 2.0 NaN
6 2 3 NaN 3.0 NaN
7 2 4 NaN 4.0 NaN
8 3 1 NaN NaN 5.0
9 3 3 NaN NaN 4.0
10 3 4 NaN NaN 3.0
11 3 5 NaN NaN 2.0
If performance is an issue, this method should be faster than this answer, especially when the number of unique values for ID increases.

How to transform weekly data to daily for specific columns using Python

I am a newbie at python and programming in general. I hope the following question is well explained.
I have a big dataset, with 80+ columns and some of these columns have only data on a weekly basis. I would like transform these columns to have values on a daily basis by simply dividing the weekly value by 7 and attributing the result to the value itself and the 6 other days of that week.
This is what my input dataset looks like:
date col1 col2 col3
02-09-2019 14 NaN 1
09-09-2019 NaN NaN 2
16-09-2019 NaN 7 3
23-09-2019 NaN NaN 4
30-09-2019 NaN NaN 5
07-10-2019 NaN NaN 6
14-10-2019 NaN NaN 7
21-10-2019 21 NaN 8
28-10-2019 NaN NaN 9
04-11-2019 NaN 14 10
11-11-2019 NaN NaN 11
..
This is what the output should look like:
date col1 col2 col3
02-09-2019 2 NaN 1
09-09-2019 2 NaN 2
16-09-2019 2 1 3
23-09-2019 2 1 4
30-09-2019 2 1 5
07-10-2019 2 1 6
14-10-2019 2 1 7
21-10-2019 3 1 8
28-10-2019 3 1 9
04-11-2019 3 2 10
11-11-2019 3 2 11
..
I can´t come up with a solution, but here is what I thought might work:
def convert_to_daily(df):
for column in df.columns.tolist():
if column.isna(): # if true
for line in range(len(df[column])):
# check if value is not empty and
succeeded by an 6 empty values or some
better logic
# I don´t know how to do that.

I believe you need select columns contains at least one missing value, forward filling missing values and divide by 7:
m = df.isna().any()
df.loc[:, m] = df.loc[:, m].ffill(limit=7).div(7)
print (df)
date col1 col2 col3
0 02-09-2019 2.0 NaN 1
1 09-09-2019 2.0 NaN 2
2 16-09-2019 2.0 1.0 3
3 23-09-2019 2.0 1.0 4
4 30-09-2019 2.0 1.0 5
5 07-10-2019 2.0 1.0 6
6 14-10-2019 2.0 1.0 7
7 21-10-2019 3.0 1.0 8
8 28-10-2019 3.0 1.0 9
9 04-11-2019 3.0 2.0 10
10 11-11-2019 3.0 2.0 11

Merging the predicted values to an existing dataframe [duplicate]

This question already has answers here:
Merge two dataframes by index
(7 answers)
Closed 2 years ago.
I'm having trouble with merging my predicted values to an existing dataframe. I currently have 2 dataframe one which has filenames and other dataframe with prediction values and both are of the same length . However when I try merging or concatenating I'm not getting a desired output.
Dataframe 1
filename
0 1gBZ9vG1.txt
1 4XztkgDw.txt
2 GfCk8XGZ.txt
3 gfHCMnJM.txt
4 GfLCd17y.txt
5 gFqruhps.txt
6 gfsZpRDu.txt
7 gfT1yDbz.txt
8 GfT9mkJL.txt
9 GFTbJDLn.txt
10 gFwh0Ekb.txt
11 GGB7680Q.txt
12 R7NkR2q2.txt
13 tK2Xmi4C.txt
Dataframe 2
predictedLabels
0 2
1 2
2 2
3 1
4 2
5 2
6 2
7 2
8 1
9 1
10 1
11 0
12 2
13 2
Output
filename predictedLabels
0 1gBZ9vG1.txt NaN
1 4XztkgDw.txt NaN
2 GfCk8XGZ.txt NaN
3 gfHCMnJM.txt NaN
4 GfLCd17y.txt NaN
5 gFqruhps.txt NaN
6 gfsZpRDu.txt NaN
7 gfT1yDbz.txt NaN
8 GfT9mkJL.txt NaN
9 GFTbJDLn.txt NaN
10 gFwh0Ekb.txt NaN
11 GGB7680Q.txt NaN
12 R7NkR2q2.txt NaN
13 tK2Xmi4C.txt NaN
0 NaN 2.0
1 NaN 2.0
2 NaN 2.0
3 NaN 1.0
4 NaN 2.0
5 NaN 2.0
6 NaN 2.0
7 NaN 2.0
8 NaN 1.0
9 NaN 1.0
10 NaN 1.0
11 NaN 0.0
12 NaN 2.0
13 NaN 2.0
I'm not sure why the labels appears below with NaN values though they are of the same length. I tried both merge and concat and also tried to reset my index but it does not work.

Try it:
Dataframe1["predictedLabels"] = Dataframe2["predictedLabels"]

Appending Pandas DataFrame in a loop

Let's say I have a df such as this:
df = pd.DataFrame({'A': [1,2,3,4,5], 'A_z': [2,3,4,5,6], 'B': [3,4,5,6,7], 'B_z': [4,5,6,7,8],
'C': [5,6,7,8,9], 'C_z': [6,7,8,9,10]})
Which looks like this:
A A_z B B_z C C_z
0 1 2 3 4 5 6
1 2 3 4 5 6 7
2 3 4 5 6 7 8
3 4 5 6 7 8 9
4 5 6 7 8 9 10
What I'm looking to do is create a new df and for each letter (A,B,C) append this new df vertically with the data from the two columns per letter so that it looks like this:
Letter Letter_z
0 1 2
1 2 3
2 3 4
3 4 5
4 5 6
5 3 4
6 4 5
7 5 6
8 6 7
9 7 8
10 5 6
11 6 7
12 7 8
13 8 9
14 9 10
As far as I'm concerned something like this should work fine:
for col in df.columns:
if col[-1] != 'z':
new_df = new_df.append(df[[col, col + '_z']])
However this results in the following mess:
A A_z B B_z C C_z
0 1.0 2.0 NaN NaN NaN NaN
1 2.0 3.0 NaN NaN NaN NaN
2 3.0 4.0 NaN NaN NaN NaN
3 4.0 5.0 NaN NaN NaN NaN
4 5.0 6.0 NaN NaN NaN NaN
0 NaN NaN 3.0 4.0 NaN NaN
1 NaN NaN 4.0 5.0 NaN NaN
2 NaN NaN 5.0 6.0 NaN NaN
3 NaN NaN 6.0 7.0 NaN NaN
4 NaN NaN 7.0 8.0 NaN NaN
0 NaN NaN NaN NaN 5.0 6.0
1 NaN NaN NaN NaN 6.0 7.0
2 NaN NaN NaN NaN 7.0 8.0
3 NaN NaN NaN NaN 8.0 9.0
4 NaN NaN NaN NaN 9.0 10.0
What am I doing wrong? Any help would be really appreciated, cheers.
EDIT:
After the kind help from jezrael the renaming of the columns in his answer got me thinking about a possible way to do it using my original train of thought.
I can now also achieve the new df I want using the following:
for col in df:
if col[-1] != 'z':
d = df[[col, col + '_z']]
d.columns = ['Letter', 'Letter_z']
new_df = new_df.append(d)
The different columns names were clearly what was causing the problem which is something I wasn't aware of at the time. Hope this helps anyone.

One ide is use Series.str.split with expand=True for MultiIndex, then use rename for avoid NaNs and finally new columns names, reshape by DataFrame.stack, sort for correct order by DataFrame.sort_index and last remove MultiIndex:
df.columns = df.columns.str.split('_', expand=True)
df = df.rename(columns=lambda x:'Letter_z' if x == 'z' else 'Letter', level=1)
df = df.stack(0).sort_index(level=[1,0]).reset_index(drop=True)
print (df)
Letter Letter_z
0 1 2
1 2 3
2 3 4
3 4 5
4 5 6
5 3 4
6 4 5
7 5 6
8 6 7
9 7 8
10 5 6
11 6 7
12 7 8
13 8 9
14 9 10
Or if possible simplify problem with reshape all non z values to one column and all z values to another use numpy.ravel:
m = df.columns.str.endswith('_z')
a = df.loc[:, ~m].to_numpy().T.ravel()
b = df.loc[:, m].to_numpy().T.ravel()
df = pd.DataFrame({'Letter': a,'Letter_z': b})
print (df)
Letter Letter_z
0 1 2
1 2 3
2 3 4
3 4 5
4 5 6
5 3 4
6 4 5
7 5 6
8 6 7
9 7 8
10 5 6
11 6 7
12 7 8
13 8 9
14 9 10

You can use the function concat and a list comprehension:
cols = df.columns[~df.columns.str.endswith('_z')]
func = lambda x: 'letter_z' if x.endswith('_z') else 'letter'
pd.concat([df.filter(like=i).rename(func, axis=1) for i in cols])
or
cols = df.columns[~df.columns.str.endswith('_z')]
pd.concat([df.filter(like=i).set_axis(['letter', 'letter_z'], axis=1, inplace=False) for i in cols])

Add Uneven Sized Data Columns in Pandas

I want to add a list as a column to the df dataframe. The list has a different size than the column length.
df =
A B C
1 2 3
5 6 9
4
6 6
8 4
2 3
4
6 6
8 4
D = [11,17,18]
I want the following output
df =
A B C D
1 2 3 11
5 6 9 17
4 18
6 6
8 4
2 3
4
6 6
8 4
I am doing the following to extend the list to the size of the dataframe by adding "nan"
# number of nan value require for the list to match the size of the column
extend_length = df.shape[0]-len(D)
# extend the list
D.extend(extend_length * ['nan'])
# add to the dataframe
df["D"] = D
A B C D
1 2 3 11
5 6 9 17
4 18
6 6 nan
8 4 nan
2 3 nan
4 nan
6 6 nan
8 4 nan
Where "nan" is treated like string but I want it to be empty ot "nan", thus, if I search for number of valid cell in D column it will provide output of 3.

Adding the list as a Series will handle this directly.
D = [11,17,18]
df.loc[:, 'D'] = pd.Series(D)

A simple pd.concat on df and series of D as follows:
pd.concat([df, pd.Series(D, name='D')], axis=1)
or
df.assign(D=pd.Series(D))
Out[654]:
A B C D
0 1 2.0 3.0 11.0
1 5 6.0 9.0 17.0
2 4 NaN NaN 18.0
3 6 NaN 6.0 NaN
4 8 NaN 4.0 NaN
5 2 NaN 3.0 NaN
6 4 NaN NaN NaN
7 6 NaN 6.0 NaN
8 8 NaN 4.0 NaN

Categories

Resources