Naive Date Compression

naive_compression.py

# 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():
        new_indexes.append(i) 

#select new df based on indexes 
compressed_df = df.loc[new_indexes].reset_index(drop=True)