Short pandas guide

Pandas

#*** Create

pandas.DataFrame([], columns=['column1','column2'])
pandas.DataFrame(data={'column1':[1,2,3], 'column2': ['1','2','3']})
pandas.DataFrame([[1,1], [2,2]], columns=['column1','column2'])

#*** Get
df.values
df.columns | df.keys()

#*** Post editing

# Rename columns
df.columns = ['column1', 'column2']

# UPDATE values by condition
df.loc[df['condition_column'] == 'condition_value', 'column_to_update'] = 0
df.loc[a['condition_column'] == 'condition_value', ['column_to_update1', 'column_to_update2']] = new_value1, new_value2
df.loc[(df['condition_column'] > 'condition_value') & (df['condition_column2'] > 'condition_value2')]

#*** Filters

# Simple value filter
df = df[df['condition_column'].isin(['1000', '-1'])]
# Simple value filter with columns select
df = df[df['condition_column'] != '-1'][['Out Peer', 'opeerid']].values


#*** Concat

# Nan value if column in child dataframe not exist
all_dfs = [pandas.DataFrame(data={'column1':[1,2,3], 'column2': ['1','2','3']}), pandas.DataFrame([[1,1], [2,2]], columns=['column1','column2'])]
new_df = pandas.concat(all_dfs)


#*** Append

# Nan value if column in child dataframe not exist
# Supports multiple rows append. In this case append method works like concat, but reassignment is necessary
new_df = df1.append(df2)


#*** Sorting
df.sort_values(by=['column1', 'column2'])

# Columns regrouping and filtering. Reassignment is necessary
new_df = df.reindex_axis(['column3','column1'], axis=1)


#*** Replace

# Replacement of null(nan) values. Reassignment is necessary
new_df = df.fillna(value='-')
new_df[['column1', 'column2']] = routelist[['column1', 'column2'']].fillna(value='NA value')

#*** Convert
df.column = df.column.astype(int, errors='ignore')