Pandas Joins, MultiIndex, Grouping and Reshaping | ISOM5650 2.3

Assemble tables with concat and joins, inspect matching multiplicity, and manipulate a named MultiIndex. Trace set_index, reset_index, sort_index, xs and IndexSlice, then compare stack, unstack, pivot, melt, pivot_table and crosstab. Learn with Professor Xuhu Wan in ISOM5650: Learning Statistics With Python. Follow real Python runs, changed inputs and interactive illustrations. First appearances of code are explained line by line; later examples focus on the new idea. RESOURCES Video + interactive slides: https://statpython.pages.dev/lesson.h... Lecture slides: https://statpython.pages.dev/section-... Course book: https://statpython-book.pages.dev/top... Course playlist:    • learnStatisticswithPython   CHAPTERS 00:00 Grouped Tables, Joins and Reshaping 00:31 Grouped Tables, Joins and Reshaping 01:03 Grouped aggregation, transform and filter 01:29 groupby: one number per group 02:15 Several aggregations at once, your own functions, named columns 03:07 transform: same shape as the input 03:52 filter: keep whole groups that pass a test 04:57 Your turn: interquartile range by weekday 06:20 Weekly resampling groups clock time, not every seventh row 07:08 Joins on multiple columns and validation 07:59 A second frame: Times university rankings 2011–2016 08:53 Joins: inner, left, outer — who survives? 09:51 Your turn: count the universities that dropped out 11:18 One research row means one security and one date 12:02 Join on both stock and date and inspect unmatched rows 13:13 Combine tables: matching columns or concatenating axes 13:41 pd.concat: side by side, or on top 14:46 Stop a join when a stock/date pair is repeated 15:59 Your turn: assemble a pair and correlate it 17:24 MultiIndex and long/wide representations 17:50 A MultiIndex identifies each row using two index levels 18:17 Six rows for the index examples 18:59 Set two columns as the index 19:59 Sort a MultiIndex before selecting a range 20:58 Move one index level into a column 21:58 Add a column to the existing index 23:03 Restore all index levels as columns 24:03 Select one stock using the outer index level 24:59 Select one date using a named index level 25:58 Select one index pair or one cell 26:52 What pd.IndexSlice means 27:18 Use IndexSlice for a range or all stocks 28:23 Swap index levels, then sort the new order 29:16 Apply the index operations to the full research panel 30:20 Reshape the return panel without aggregating repeated stock/date pairs 30:48 Run unstack and compare the Series with the table 31:47 Stack columns into an inner index level 32:34 Missing combinations after unstack 33:38 Transpose swaps the entire row and column axes 34:36 Verify the full long–wide–long round trip 35:48 melt, pivot and pivot_table: the structural difference 36:17 pivot: choose the row index, column headings and values 37:18 melt: retain identifiers and turn value columns into rows 38:25 pivot_table: two grouping columns become rows × columns 38:53 Run pivot_table with explicit keyword arguments 39:57 The same table by groupby + unstack 40:24 Inspect the grouped Series before unstacking it 41:27 Apply the comparison to four countries and six years 42:56 melt and pivot on the five-stock price panel 44:07 Inventory prices by release year: pivot_table and crosstab 45:08 Inventory counts: unstack either named level 46:09 Inventory prices: stack flag columns into the index 47:11 Inventory prices: transpose the year and flag axes 48:10 pd.cut: assign values to intervals 49:02 Plotting a frame 49:48 Your turn: citations vs rank, top 100 of 2016