Pandas Essentials Study Guide: 93 Flashcards on DataFrames, Indexing, GroupBy and Merge

Most people don't struggle with Pandas because the concepts are hard. They struggle because they keep reopening the documentation for the same twenty methods: is it loc or iloc, does drop take axis=1 for columns, is it isnull or isna? The Pandas Essentials deck on Flip It (code VX1180) turns those 93 recurring questions into flashcards so the answers come from memory instead of from a browser tab.

What the 93 cards cover

The deck reads like a condensed version of the "Python for Data Analysis" essentials, from the two core data structures through to time series and aggregation. Fronts are either a method name or a task ("Filter rows based on a condition"), backs are the exact syntax or a one-line explanation.

Grouped by theme, you will memorize:

• Basics: what Pandas is (built on NumPy, written in Python, Cython and C), the origin of the name (panel data), import pandas as pd, Series vs DataFrame, default integer index, index alignment when adding two Series.
• Inspecting data: head(), describe(), info(), shape, dtypes, columns, index, values, empty, T, nunique(), value_counts(), sample(n), nlargest(), nsmallest().
• Selecting and filtering: column access as a Series, loc vs iloc, multiple columns, boolean conditions with & and |, sort_values(), idxmax() / idxmin().
• Cleaning: NaN and NaT, isnull() / isna() / notnull(), fillna(), dropna(), duplicated(), drop_duplicates(), astype(), rename(), drop(), copy(), reset_index(drop=True), set_index().
• Combining and reshaping: merge() with on, left_on / right_on, concat(), axis=0 vs axis=1, pivot_table(), melt(), MultiIndex.
• Aggregation: groupby(), mean(), sum(), median(), std(), corr(), agg(), apply(), pd.cut().
• Time series: pd.date_range(), slicing by date string, resample(), the 'W-Mon' frequency alias.
• I/O and performance: read_csv, read_excel, to_csv, supported formats (CSV, JSON, Parquet, SQL, Excel), Wes McKinney's 5-10× RAM rule of thumb and his recommendation of Apache Arrow.

How to study syntax cards without cheating yourself

Code flashcards fail when you "recognise" the answer instead of producing it. Two habits fix that:

1. Type the back before you flip. Keep a Python REPL or a Jupyter cell open next to Flip It. For every card whose front is a task ("Set a column as the index", "Select rows by multiple conditions"), type your answer on a real DataFrame first, then flip. If your line runs and matches, mark it known. If you wrote df[df['a'] > 5 and df['b'] < 10] and got a ValueError, you have just learned the & card for good.

2. Drill the confusable pairs back to back. The deck deliberately contains near-twins: loc / iloc, isnull / isna / notnull, merge / concat, axis=0 / axis=1, pivot_table / melt, idxmax / nlargest. When one of them comes up in review, say the other one's definition too. Mixing them is the single most common Pandas interview slip.

A two-week schedule

• Days 1-4: 15 new cards a day, in deck order (basics → inspection → selection → cleaning).
• Days 5-7: the remaining 33 cards (combining, aggregation, time series, I/O).
• Days 8-14: reviews only, plus one small dataset per day where you apply at least five cards from memory.

At that pace the whole deck takes under 15 minutes a day.

Related decks

If your Python fundamentals are shaky, pair this deck with our guide to Python data structures and comprehensions; for the SQL side of merge() and groupby(), see how to master SQL queries, joins and aggregates.

Log in to Flip It, search for deck code VX1180 — Pandas Essentials, and start with the first 15 cards today.