Revision history [back]

One way to resample the pandas dataframe using multiple groupbys to ensure that each condition has an equal number of days worth of data is as follows:

1. Group the dataframe by the condition(s) and date, and count the number of rows for each group.
2. Calculate the minimum count value for all groups.
3. Filter out rows for each group that exceed the minimum count value.
4. Concatenate the filtered groups back into a single dataframe.

Assuming the conditions are in columns 'condition1' and 'condition2', and the dates are in column 'date', the code to implement the above steps would look like:

# Step 1: Group by condition(s) and date, and count the number of rows
grouped = df.groupby(['condition1', 'condition2', 'date']).count()

# Step 2: Calculate the minimum count value for all groups
min_count = grouped.groupby(['condition1', 'condition2']).min()['col_name']

# Step 3: Filter out rows for each group that exceed the minimum count value
filtered_groups = []
for group_name, group_data in grouped:
    if group_data.shape[0] > min_count.loc[group_name[0], group_name[1]]:
        filtered_groups.append(group_data.sample(min_count.loc[group_name[0], group_name[1]]))
    else:
        filtered_groups.append(group_data)

# Step 4: Concatenate the filtered groups back into a single dataframe
filtered_df = pd.concat(filtered_groups)

Note that 'col_name' in Step 2 refers to any column in the dataframe that can be counted to determine the number of rows in each group. This column can be replaced with any other column that has no missing or null values.