Created
November 21, 2023 02:59
-
-
Save royerk/0f795c2dfadc20744a86dffac445105a to your computer and use it in GitHub Desktop.
AB infer missing values, pick first value from day before
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
import pandas as pd | |
import numpy as np | |
df = pd.DataFrame( | |
[ | |
["id_0", "2023-11-01", np.nan, 1, 0.5], | |
["id_0", "2023-11-02", 0.5, 2, np.nan], | |
["id_0", "2023-11-03", np.nan, np.nan, np.nan], # device in baseline | |
["id_0", "2023-11-04", np.nan, np.nan, np.nan], # device in baseline | |
["id_0", "2023-11-05", np.nan, 1.0, np.nan], | |
], | |
columns=["id", "date", "0", "1", "2"], | |
) | |
df["date"] = pd.to_datetime(df["date"]) | |
print(df) | |
expected_df = pd.DataFrame( | |
[ | |
["id_0", "2023-11-01", np.nan, 1, 0.5], | |
["id_0", "2023-11-02", 0.5, 2.0, 2.0], | |
["id_0", "2023-11-03", 2.0, 2.0, 2.0], # device in baseline | |
["id_0", "2023-11-04", 2.0, 2.0, 2.0], # device in baseline | |
["id_0", "2023-11-05", 2.0, 1.0, 1.0], | |
], | |
columns=["id", "date", "0", "1", "2"], | |
) | |
expected_df["date"] = pd.to_datetime(expected_df["date"]) | |
def find_last_value(row, columns): | |
non_null_values = row[columns][row.notnull()] | |
if len(non_null_values) > 0: | |
return non_null_values[-1] | |
return np.nan | |
def fill_row_set_points(row, columns): | |
last_value = row["first_value"] | |
for i in columns: | |
if pd.isnull(row[i]): | |
row[i] = last_value | |
else: | |
last_value = row[i] | |
return row | |
def fill_df_set_points(df, columns): | |
df["last_value"] = df.apply(lambda x: find_last_value(x, columns), axis=1) | |
df_join = df[["id", "next_date", "last_value"]].copy() | |
df_join.rename( | |
columns={"next_date": "date", "last_value": "first_value"}, inplace=True | |
) | |
df_join.set_index(["id", "date"], inplace=True) | |
df.set_index(["id", "date"], inplace=True) | |
df.drop(columns=["last_value"], inplace=True) | |
df = pd.merge(df, df_join, how="left", left_index=True, right_index=True) | |
df.reset_index(inplace=True) | |
df = df.apply(fill_row_set_points, axis=1, columns=columns) | |
df.drop(columns=["first_value"], inplace=True) | |
return df | |
def populate_set_points(df): | |
columns = ["0", "1", "2"] # todo: get columns from df | |
missing_values = -1 | |
df["next_date"] = df["date"].apply(lambda x: x + pd.DateOffset(days=1)) | |
iteration = 0 | |
while missing_values != df[columns].isnull().sum().sum(): | |
print(f"iteration {iteration}") | |
iteration += 1 | |
missing_values = df[columns].isnull().sum().sum() | |
df = fill_df_set_points(df) | |
df.drop(columns=["next_date"], inplace=True) | |
return df | |
df = populate_set_points(df) | |
print(df) | |
pd.testing.assert_frame_equal(df, expected_df) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment