classification_crypto_bal_data.py

def balanceDecisions(n_df):
    """
    Rebalances data so that each class/decision is represented equally.
    """
    # Counting each class or decision
    counts = n_df['decision'].value_counts()

    # The lowest number of represented classes
    low_num = counts.min()

    # Sampling from the DF with the latest data
    df1 = n_df[n_df['decision']==counts.sort_values().index[0]].tail(low_num)

    df2 = n_df[n_df['decision']==counts.sort_values().index[1]].tail(low_num)

    # Combining the resampled DFs and returning it
    return df1.append(df2)