Python & pandas 101: the formulas

Python in one breath

x = 5                                   # variable
stages = ['qualified', 'proposal']      # list; stages[0], len(stages)
deal = {'id': 'D-1', 'amount': 5000}    # dict; deal['amount']
round(2 / 3, 4)                         # 0.6667
[s for s in stages if s != 'proposal']  # list comprehension (a filter)

== compares, = assigns. // is integer division. Indentation defines blocks.

pandas ↔ SQL

SQL	pandas
SELECT deal_id, amount FROM deals	deals[['deal_id', 'amount']]
WHERE stage = 'closed_won'	deals[deals['stage'] == 'closed_won']
WHERE a AND b	deals[(cond_a) & (cond_b)] — parentheses required
WHERE stage IN (...)	deals[deals['stage'].isin([...])]
COUNT(*)	len(deals)
SUM(amount)	deals['amount'].sum()
GROUP BY owner, SUM(amount)	deals.groupby('owner')['amount'].sum()
ORDER BY amount DESC	.sort_values('amount', ascending=False)
LIMIT 5	.head(5)
JOIN accounts ON account_id	deals.merge(accounts, on='account_id')
LEFT JOIN	.merge(..., how='left')
several aggregates	.agg(total=('amount', 'sum'), n=('deal_id', 'count'))

Getting an answer out

result = df.to_dict('records') for rows; int(x) / float(x) to turn a numpy number into a plain one; (deals['stage'] == 'closed_won').sum() counts True values, the pandas CASE WHEN.