gautierdag

cols = [d for d in df.columns if d != "date"]
compressed_df = df.loc[~(df[cols].shift() == df[cols]).all(axis=1)].reset_index(drop=True)

gautierdag

from functools import reduce

values = df.values
new_values = []
cols = [df.columns.get_loc(d) for d in df.columns if d != "date"]
new_values.append(values[0])

def compress_helper(x, y):
    if (x[cols] == y[cols]).all():
        return x

gautierdag

# set first (oldest balance as index entry)
new_indexes = [0]
# choose columns that are not the date column
cols = [d for d in df.columns if d != "date"]
# iterate over rows
for i, row in df.iterrows():
    if i == 0: #skip 0 index since it has no previous balance
        continue
    # if not all values of the previous row/date is not equal to current - then we have new observation
    if not (df.iloc[i-1][cols] ==row[cols]).all():

gautierdag

from random import randint
import pandas as pd

num_observations = 500
dates = pd.to_datetime([f"{randint(2018, 2020)}-{randint(1, 12)}-{randint(1, 27)}" 
                        for _ in range(num_observations)]).unique()
df = pd.DataFrame({"date": dates, 
                   "balance": [randint(0, 10000) 
                               for _ in range(len(dates))]
                  })

gautierdag

import datetime
from typing import Optional, List
from pydantic import BaseModel


class PlaidBalance(BaseModel):
    current: float
    available: Optional[float]
    iso_currency_code: Optional[str]
    limit: Optional[float]