We Keep Coding

pandas dataframe interpolate for Nans with groupby using window of discrete days of the year

The small reproducible example below sets up a dataframe that is 100 yrs in length containing some randomly generated values. It then inserts 3 100-day stretches of missing values. Using this small example, I am attempting to sort out the pandas commands that will fill in the missing days using average values for that day of the year (hence the use of .groupby) with a condition. For example, if April 12th is missing, how can the last line of code be altered such that only the 10 nearest April 12th's are used to fill in the missing value? In other words, a missing April 12th value in 1920 would be filled in using the mean April 12th values between 1915 to 1925; a missing April 12th value in 2000 would be filled in with the mean April 12th values between 1995 to 2005, etc. I tried playing around with adding a .rolling() to the lambda function in last line of script, but was unsuccessful in my attempt.
Bonus question: The example below extends from 1918 to 2018. If a value is missing on April 12th 1919, for example, it would still be nice if ten April 12ths were used to fill in the missing value even though the window couldn't be 'centered' on the missing day because of its proximity to the beginning of the time series. Is there a solution to the first question above that would be flexible enough to still use a minimum of 10 values when missing values are close to the beginning and ending of the time series?
import pandas as pd
import numpy as np
import random
# create 100 yr time series
dates = pd.date_range(start="1918-01-01", end="2018-12-31").strftime("%Y-%m-%d")
vals = [random.randrange(1, 50, 1) for i in range(len(dates))]
# Create some arbitrary gaps
vals[100:200] = vals[9962:10062] = vals[35895:35995] = [np.nan] * 100
# Create dataframe
df = pd.DataFrame(dict(
list(
zip(["Date", "vals"],
[dates, vals])
)
))
# confirm missing vals
df.iloc[95:105]
df.iloc[35890:35900]
# set a date index (for use by groupby)
df.index = pd.DatetimeIndex(df['Date'])
df['Date'] = df.index
# Need help restricting the mean to the 10 nearest same-days-of-the-year:
df['vals'] = df.groupby([df.index.month, df.index.day])['vals'].transform(lambda x: x.fillna(x.mean()))

This answers both parts
build a DF dfr that is the calculation you want
lambda function returns a dict {year:val, ...}
make sure indexes are named in reasonable way
expand out dict with apply(pd.Series)
reshape by putting year columns back into index
merge() built DF with original DF. vals column contains NaN 0 column is value to fill
finally fillna()
# create 100 yr time series
dates = pd.date_range(start="1918-01-01", end="2018-12-31")
vals = [random.randrange(1, 50, 1) for i in range(len(dates))]
# Create some arbitrary gaps
vals[100:200] = vals[9962:10062] = vals[35895:35995] = [np.nan] * 100
# Create dataframe - simplified from question...
df = pd.DataFrame({"Date":dates,"vals":vals})
df[df.isna().any(axis=1)]
ystart = df.Date.dt.year.min()
# generate rolling means for month/day. bfill for when it's start of series
dfr = (df.groupby([df.Date.dt.month, df.Date.dt.day])["vals"]
.agg(lambda s: {y+ystart:v for y,v in enumerate(s.dropna().rolling(5).mean().bfill())})
.to_frame().rename_axis(["month","day"])
)
# expand dict into columns and reshape to by indexed by month,day,year
dfr = dfr.join(dfr.vals.apply(pd.Series)).drop(columns="vals").rename_axis("year",axis=1).stack().to_frame()
# get df index back, plus vals & fillna (column 0) can be seen alongside each other
dfm = df.merge(dfr, left_on=[df.Date.dt.month,df.Date.dt.day,df.Date.dt.year], right_index=True)
# finally what we really want to do - fill tha NaNs
df.fillna(dfm[0])
analysis
taking NaN for 11-Apr-1918, default is 22 as it's backfilled from 1921
(12+2+47+47+2)/5 == 22
dfm.query("key_0==4 & key_1==11").head(7)
key_0
key_1
key_2
Date
vals
0
100
4
11
1918
1918-04-11 00:00:00
nan
22
465
4
11
1919
1919-04-11 00:00:00
12
22
831
4
11
1920
1920-04-11 00:00:00
2
22
1196
4
11
1921
1921-04-11 00:00:00
47
27
1561
4
11
1922
1922-04-11 00:00:00
47
36
1926
4
11
1923
1923-04-11 00:00:00
2
34.6
2292
4
11
1924
1924-04-11 00:00:00
37
29.4

I'm not sure how far I've gotten with the intent of your question. The approach I've taken is to satisfy two requirements
Need an arbitrary number of averages
Use those averages to fill in the NA
I have addressed the
Simply put, instead of filling in the NA with before and after dates, I fill in the NA with averages extracted from any number of years in a row.
import pandas as pd
import numpy as np
import random
# create 100 yr time series
dates = pd.date_range(start="1918-01-01", end="2018-12-31").strftime("%Y-%m-%d")
vals = [random.randrange(1, 50, 1) for i in range(len(dates))]
# Create some arbitrary gaps
vals[100:200] = vals[9962:10062] = vals[35895:35995] = [np.nan] * 100
# Create dataframe
df = pd.DataFrame(dict(
list(
zip(["Date", "vals"],
[dates, vals])
)
))
df['Date'] = pd.to_datetime(df['Date'])
df['mm-dd'] = df['Date'].apply(lambda x:'{:02}-{:02}'.format(x.month, x.day))
df['yyyy'] = df['Date'].apply(lambda x:'{:04}'.format(x.year))
df = df.iloc[:,1:].pivot(index='mm-dd', columns='yyyy')
df.columns = df.columns.droplevel(0)
df['nans'] = df.isnull().sum(axis=1)
df['10n_mean'] = df.iloc[:,:-1].sample(n=10, axis=1).mean(axis=1)
df['10n_mean'] = df['10n_mean'].round(1)
df.loc[df['nans'] >= 1]
yyyy 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 ... 2011 2012 2013 2014 2015 2016 2017 2018 nans 10n_mean
mm-dd
02-29 NaN NaN 34.0 NaN NaN NaN 2.0 NaN NaN NaN ... NaN 49.0 NaN NaN NaN 32.0 NaN NaN 76 21.6
04-11 NaN 43.0 12.0 28.0 29.0 28.0 1.0 38.0 11.0 3.0 ... 17.0 35.0 8.0 17.0 34.0 NaN 5.0 33.0 3 29.7
04-12 NaN 19.0 38.0 34.0 48.0 46.0 28.0 29.0 29.0 14.0 ... 41.0 16.0 9.0 39.0 8.0 NaN 1.0 12.0 3 21.3
04-13 NaN 33.0 26.0 47.0 21.0 26.0 20.0 16.0 11.0 7.0 ... 5.0 11.0 34.0 28.0 27.0 NaN 2.0 46.0 3 21.3
04-14 NaN 36.0 19.0 6.0 45.0 41.0 24.0 39.0 1.0 11.0 ... 30.0 47.0 45.0 14.0 48.0 NaN 16.0 8.0 3 24.7
df_mean = df.T.fillna(df['10n_mean'], downcast='infer').T
df_mean.loc[df_mean['nans'] >= 1]
yyyy 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 ... 2011 2012 2013 2014 2015 2016 2017 2018 nans 10n_mean
mm-dd
02-29 21.6 21.6 34.0 21.6 21.6 21.6 2.0 21.6 21.6 21.6 ... 21.6 49.0 21.6 21.6 21.6 32.0 21.6 21.6 76.0 21.6
04-11 29.7 43.0 12.0 28.0 29.0 28.0 1.0 38.0 11.0 3.0 ... 17.0 35.0 8.0 17.0 34.0 29.7 5.0 33.0 3.0 29.7
04-12 21.3 19.0 38.0 34.0 48.0 46.0 28.0 29.0 29.0 14.0 ... 41.0 16.0 9.0 39.0 8.0 21.3 1.0 12.0 3.0 21.3
04-13 21.3 33.0 26.0 47.0 21.0 26.0 20.0 16.0 11.0 7.0 ... 5.0 11.0 34.0 28.0 27.0 21.3 2.0 46.0 3.0 21.3
04-14 24.7 36.0 19.0 6.0 45.0 41.0 24.0 39.0 1.0 11.0 ... 30.0 47.0 45.0 14.0 48.0 24.7 16.0 8.0 3.0 24.7

View data after transformation

Is it possible to see the data after Altair applies transformations and aggregations?
For example, can you access the underlying data after the following transformations?
import altair as alt
from vega_datasets import data
source = data.seattle_weather.url
step = 20
overlap = 1
alt.Chart(source, height=step).transform_timeunit(
Month='month(date)'
).transform_joinaggregate(
mean_temp='mean(temp_max)', groupby=['Month']
).transform_bin(
['bin_max', 'bin_min'], 'temp_max'
).transform_aggregate(
value='count()', groupby=['Month', 'mean_temp', 'bin_min', 'bin_max']
).transform_impute(
impute='value', groupby=['Month', 'mean_temp'], key='bin_min', value=0
).mark_area(...
)
Above code from the Ridgeplot example

Transforms are evaluated in Javascript, and there is not any built-in way to access data in the javascript frontend from the Python backend. However, there is an experimental package called altair_transform that is able to evaluate most Vega expressions in Python.
For your chart you can use it like this:
import altair as alt
from vega_datasets import data
source = data.seattle_weather()
step = 20
overlap = 1
chart = alt.Chart(source, height=step).transform_timeunit(
Month='month(date)'
).transform_joinaggregate(
mean_temp='mean(temp_max)', groupby=['Month']
).transform_bin(
['bin_max', 'bin_min'], 'temp_max'
).transform_aggregate(
value='count()', groupby=['Month', 'mean_temp', 'bin_min', 'bin_max']
).transform_impute(
impute='value', groupby=['Month', 'mean_temp'], key='bin_min', value=0
).mark_area().encode(
x='Month:T',
y='value:Q'
)
import altair_transform
data = altair_transform.extract_data(chart)
print(data)
bin_min Month mean_temp bin_max value
0 0.0 1900-01-01 8.229032 -5.0 2.0
1 5.0 1900-01-01 8.229032 0.0 19.0
2 10.0 1900-01-01 8.229032 5.0 72.0
3 15.0 1900-01-01 8.229032 10.0 29.0
4 20.0 1900-01-01 8.229032 15.0 2.0
.. ... ... ... ... ...
103 20.0 1900-12-01 8.194355 15.0 4.0
104 25.0 1900-12-01 8.194355 NaN 0.0
105 30.0 1900-12-01 8.194355 NaN 0.0
106 35.0 1900-12-01 8.194355 NaN 0.0
107 40.0 1900-12-01 8.194355 NaN 0.0
[108 rows x 5 columns]

Python: Array-based equation

I have a dataframe 500 rows long by 4 columns. I need to find the proper python code that would divide the current row by the row below and then multiply that value by the value in the last row for every value in each column. I need to replicate this excel formula basically.

It's not clear if your data is stored in an array as provided by Numpy, were it true you'd write, with the original data contained in a
b = a[-1]*(a[:-1]/a[+1:])
a[-1] is the last row, a[:-1] the array without the last row and a[+1:] the array without the first (index zero, that is) row

Assuming you are talking about pandas DataFrame
import pandas as pd
import random
# sample DataFrame object
df = pd.DataFrame((float(random.randint(1, 100)),
float(random.randint(1, 100)),
float(random.randint(1, 100)),
float(random.randint(1, 100)))
for _ in range(10))
def function(col):
for i in range(len(col)-1):
col[i] = (col[i]/col[i+1])*col[len(col)-1]
print(df) # before formula apply
df.apply(function)
print(df) # after formula apply
>>>
0 1 2 3
0 10.0 78.0 27.0 23.0
1 72.0 42.0 77.0 86.0
2 82.0 12.0 58.0 98.0
3 27.0 92.0 19.0 86.0
4 48.0 83.0 14.0 43.0
5 55.0 18.0 58.0 77.0
6 20.0 58.0 20.0 22.0
7 76.0 19.0 63.0 82.0
8 23.0 99.0 58.0 15.0
9 60.0 57.0 89.0 100.0
0 1 2 3
0 8.333333 105.857143 31.207792 26.744186
1 52.682927 199.500000 118.155172 87.755102
2 182.222222 7.434783 271.684211 113.953488
3 33.750000 63.180723 120.785714 200.000000
4 52.363636 262.833333 21.482759 55.844156
5 165.000000 17.689655 258.100000 350.000000
6 15.789474 174.000000 28.253968 26.829268
7 198.260870 10.939394 96.672414 546.666667
8 23.000000 99.000000 58.000000 15.000000
9 60.000000 57.000000 89.000000 100.000000

Plotting multiple lines, want a mark every with a text label on each line

I am currently working on a graph where I have 4 lines going through it. I want to have a bunch of marks on it so that people can get a specific value at that point and time on the graph.
Attached is my example. I have a bunch of lines with various values. I would like to set it up to where every 30 days, it places with a mark on the line with the value of that point. That way people can more easily tell the value.
Example of my dataset
25 50 90 100
2019-04-04 55.0 76.0 1027.0 1200.0
2019-04-05 56.0 77.0 1028.0 1201.0
2019-04-06 57.0 78.0 1029.0 1202.0
2019-04-07 58.0 79.0 1030.0 1203.0
2019-04-08 59.0 80.0 1031.0 1204.0
2019-04-09 60.0 81.0 1032.0 1205.0
2019-04-10 61.0 82.0 1033.0 1206.0
2019-04-11 62.0 83.0 1034.0 1207.0
2019-04-12 53.0 84.0 1035.0 1208.0
2019-04-13 54.0 85.0 1036.0 1209.0
2019-04-14 55.0 86.0 1037.0 1210.0
2019-04-15 56.0 87.0 1038.0 1211.0
2019-04-16 57.0 88.0 1039.0 1212.0
2019-04-17 58.0 89.0 1040.0 1213.0
2019-04-18 59.0 90.0 1041.0 1214.0
2019-04-19 60.0 91.0 1042.0 1215.0
2019-04-20 61.0 92.0 1043.0 1216.0
2019-04-21 62.0 93.0 1044.0 1217.0
2019-04-22 63.0 94.0 1045.0 1218.0
2019-04-23 64.0 95.0 1046.0 1219.0
2019-04-24 65.0 96.0 1047.0 1220.0
2019-04-25 66.0 97.0 1048.0 1221.0
2019-04-26 67.0 98.0 1049.0 1222.0
2019-04-27 68.0 99.0 1050.0 1223.0
2019-04-28 69.0 100.0 1051.0 1224.0
2019-04-29 70.0 101.0 1052.0 1225.0
2019-04-30 71.0 102.0 1053.0 1226.0
2019-05-01 72.0 103.0 1054.0 1227.0
2019-05-02 73.0 104.0 1055.0 1228.0
2019-05-03 74.0 105.0 1056.0 1229.0
And the code I am using to plot
plt.rcParams['figure.figsize'] = [18, 10]
df = pd.DataFrame(data=panda_data)
fig, ax = plt.subplots()
ax = df.plot(kind='line')
ax.grid(axis='y')

With daily data, use resample.asfreq() to get the points every X days (I'll use 15 here for illustration given the date ranges). Then plot them and annotate. Ensuring the labels don't overlap is a much harder task.
For a few points, don't worry too much about the performance of stack + iteritems.
import matplotlib.pyplot as plt
import pandas as pd
#df.index = pd.to_datetime(df.index) # If not a DatetimeIndex
pts = df.resample('15D').asfreq()
fig, ax = plt.subplots(figsize=(10, 6))
df.plot(kind='line', ax=ax)
pts.plot(marker='o', ax=ax, lw=0, color='black', legend=False)
pad = 10
for idx, val in pts.stack().iteritems():
ax.annotate(val, (idx[0], val+pad))
ax.grid(axis='y')

Creating evenly spaced time intervals while interpolating values from original data

I have a csv file that is formatted like this:
Time | RH
----------------
1491062400| 39.0
1491062590| 40.0
1491062891| 40.0
1491063191| 39.0
1491063490| 40.0
1491063791| 40.0
1491064090| 39.0
1491064390| 38.0
1491064689| 39.0
1491064990| 40.0
1491065289| 39.0
1491065590| 39.0
1491065890| 40.0
1491022991| 39.0
1491023292| 39.0
1491023592| 38.0
1491023893| 39.0
1491024194| 39.0
1491024494| 39.0
1491024795| 40.0
1491025094| 40.0
1491025395| 39.0
1491025695| 40.0
1491025995| 40.0
1491026295| 39.0
1491026594| 39.0
1491026895| 37.0
1491027195| 39.0
1491027494| 39.0
But the time intervals, as seen, jumps randomly (i.e a 3 minute jump, 10 minute jump, etc.) I need to figure out a way to evenly space this time data into 5 minute intervals while interpolating the value under 'RH' to keep everything accurate.