4D Training & Consultancy

Software Development

Data Wrangling with Python and pandas

A hands-on pandas course for teams that spend more time repairing data than analysing it. Participants work through dtypes and memory, vectorized operations, merge and join semantics, missing-value strategy, reshaping, and time-series alignment, ending with cleaning code that runs unchanged next month.

3 daysIn-house, online, or customized deliveryCorporate teams and professional groupsLevel: Foundation to Intermediate

Overview

Practical learning for workplace transfer.

The common failure mode is not a missing feature in pandas — it is a notebook that worked once and breaks the moment the source file changes. A column arrives as text instead of a number, a left join silently multiplies rows, a chained assignment updates a copy nobody sees, and the reported total quietly drifts. This course rebuilds the foundations that prevent all four: what an Index actually guarantees, why dtype selection decides both correctness and memory, how merge validation catches duplicate keys before they reach a report, and how to express a transformation without a Python loop. Participants finish by refactoring a recurring cleaning task into a tested script.

Prerequisites

Basic Python syntax including functions, lists, and dictionaries. No prior pandas experience is required.

Objectives

  • Select dtypes deliberately so datasets stay correct and fit comfortably in memory.
  • Replace row-by-row loops and apply calls with vectorized pandas operations.
  • Apply merge, join, and concat with validated keys and predictable row counts.
  • Choose a defensible missing-data strategy and document its effect on results.
  • Reshape between wide and long form using pivot, melt, stack, and groupby.
  • Package a cleaning workflow as a parameterised script with data-quality checks.

Target audience

  • Data analysts moving from spreadsheets to Python
  • Business intelligence and reporting specialists
  • Finance, planning, and commercial analysts handling recurring data files
  • Junior data engineers building ingestion and cleaning routines
  • Researchers and scientists preparing experimental or field datasets
  • Operations and quality teams consolidating data from multiple systems

Program outline

A clear structure for the learning journey.

Program outline

Outline points are grouped in one designed block instead of being treated as separate module cards.

Module 1: Series, DataFrames, dtypes, and Memory

The Index contract: alignment, duplicates, and why operations silently reorder

int, float, object, category, and nullable dtypes compared on real columns

Measuring memory with memory_usage and cutting it with category and downcasting

Views versus copies, chained indexing, and the SettingWithCopy trap

Module 2: Loading Messy Sources and Selecting Data

read_csv, read_excel, and JSON options for encodings, separators, and bad rows

Declaring dtype and parse_dates at load time instead of repairing afterwards

loc, iloc, boolean masks, query, and when each is the readable choice

Reading files larger than memory with chunksize and column selection

Module 3: Cleaning, Missing Data, and Type Correction

NaN, None, NaT, and pd.NA: how each propagates through arithmetic and comparison

Drop, fill, forward-fill, and interpolate: choosing by business meaning, not habit

String cleaning with the .str accessor, regex extraction, and normalisation

Detecting duplicates, outliers, and impossible values before they reach a report

Module 4: Combining Datasets Without Losing Rows

Inner, left, right, and outer joins and the row count each one implies

The validate argument as a contract against many-to-many key explosions

indicator, suffixes, and reconciling columns that exist on both sides

concat, merge_asof, and combining files that share only part of a schema

Module 5: Grouping, Reshaping, and Time Series

groupby split-apply-combine with agg, transform, and filter

pivot_table, melt, stack, and unstack for wide and long reshaping

DatetimeIndex, time zones, resample, and rolling windows

Window functions for running totals, ranks, and period-over-period change

Module 6: From Notebook to Repeatable Cleaning Pipeline

Refactoring notebook cells into named, tested transformation functions

Assertion-based data-quality gates on row counts, ranges, and key uniqueness

Parameterising paths, dates, and filters so the script runs unattended

Writing to Parquet and CSV with stable schemas and audit columns

Materials provided

  • Course workbook, annotated code samples, and reference notes
  • Hands-on lab environment and starter repositories
  • Exercises, checklists, and reusable code templates
  • 4D Certificate of Completion
  • Post-course technical guidance

Training Options

Programs can be delivered in-house, online, or in a blended format depending on your team's schedule, location, and learning objectives. When an external certificate or exam is included, certification rules and fees remain under the relevant awarding body's policies, while 4D provides the training and preparation support.

Why choose 4D

4D runs this course on your own extracts — the ERP export with mixed date formats, the monthly file whose headers move, the join that keeps inflating totals. Trainers profile those files live, rebuild the cleaning logic in pandas alongside the analysts who own it, and leave the team with a tested script that replaces the manual repair step in their reporting cycle.

Speak to 4D

Plan the right training or consultancy path for your team.

Share a few details and 4D will help route your inquiry toward corporate training, consultancy, assessment, Phoenix-enabled support, or a tailored program.