Data Analytics with Pandas Training Course

Participants can import common source formats and control how Pandas interprets incoming data.

• CSV imports with delimiters, encodings, headers and missing-value markers
• Excel workbook imports using sheet names, ranges and column selection
• JSON records normalised into tabular DataFrames with json_normalize
• SQL query results loaded through database connection interfaces
• Specifying dtypes during import to prevent incorrect inference
• Import audit notes recording source, extraction date and row counts

Participants can assess dataset condition and document material data-quality risks.

• DataFrame info, describe and value_counts for initial data profiling
• Null counts, uniqueness rates and cardinality checks by column
• Frequency tables revealing unexpected categories and spelling variants
• Range checks against valid dates, amounts and operational thresholds
• Duplicate-record detection using business keys and row comparisons
• Data-quality issue log capturing severity, owner and treatment decision

Participants can retrieve the correct records and columns using explicit selection logic.

• Column selection with brackets, loc and positional iloc indexing
• Boolean masks for multi-condition filtering and exclusion rules
• Query method syntax for readable row-selection expressions
• Setting indexes for labelled selection and alignment operations
• Sorting records by multiple fields, direction and null placement
• Sampling rows to inspect subsets without changing source data

Participants can standardise text fields and construct controlled analytical categories.

• String methods for trimming whitespace, case normalisation and replacement
• Regular expressions for extracting codes, dates and structured text
• Category mappings converting inconsistent labels into reporting classifications
• Split and extract methods for compound identifiers and addresses
• Detecting near-duplicate categories through frequency and pattern review
• Documenting category rules in a maintained reference mapping table

Participants can treat missing and invalid values using stated analytical rules.

• isna and notna methods for locating absent values
• Missingness patterns assessed by column, segment and reporting period
• dropna criteria matched to analysis purpose and business risk
• fillna strategies using constants, groups, forward-fill and backfill
• Invalid values replaced after rule-based detection and evidence review
• Treatment decisions recorded for audit and stakeholder review

Participants can assign appropriate data types and avoid common calculation errors.

• astype conversions for strings, integers, floats, booleans and categories
• to_numeric conversion with coercion rules for malformed numeric fields
• Decimal considerations for currency calculations and financial reporting
• Integer nullability managed with Pandas nullable integer data types
• Memory usage reduced through suitable categorical and numeric dtypes
• Type-validation checks applied before aggregations and business calculations

Participants can derive analytical fields from source data using transparent transformation logic.

• Vectorised arithmetic for margins, rates, variances and unit measures
• where and mask methods for conditional value replacement
• np.select rules for multi-condition classification and status assignment
• assign method for readable chained column transformations
• apply method limits compared with vectorised Pandas operations
• Business-rule register linking derived fields to calculation definitions

Participants can calculate grouped measures that answer defined business questions.

• groupby operations by region, product, customer and reporting period
• Named aggregations for totals, means, counts, minima and maxima
• Multiple aggregation functions applied to selected numerical columns
• transform method for group-relative measures and benchmark comparisons
• Ranking groups using rank, sorting and tie-handling methods
• Aggregation validation against source totals and control samples

Participants can convert datasets between wide and long forms for analysis and presentation.

• pivot_table summaries with values, indexes, columns and aggregation functions
• pivot method requirements for unique index-column combinations
• melt transformations converting repeated measure columns into long format
• stack and unstack operations for hierarchical index restructuring
• crosstab tables for categorical counts, proportions and margins
• Wide-versus-long design choices for charts, models and reports

Participants can combine datasets safely and test whether records joined as intended.

• merge operations using inner, left, right and outer joins
• Join-key preparation through type matching, trimming and standardisation
• validate parameter checks for one-to-one and one-to-many relationships
• Merge indicators identifying unmatched records from either source
• concat operations for appending aligned monthly or regional extracts
• Reconciliation report comparing pre-merge and post-merge record counts

Participants can convert, extract and analyse dates for operational and reporting purposes.

• to_datetime parsing with formats, errors handling and timezone awareness
• Datetime accessors extracting year, quarter, month, weekday and hour
• Period types for monthly, quarterly and financial-year reporting
• Date offsets calculating due dates, ageing bands and elapsed periods
• Business-day calendars excluding weekends and defined public holidays
• Date-quality checks for impossible sequences and future-dated records

Participants can produce time-based measures for trends, comparisons and monitoring.

• Datetime indexes prepared for resampling and chronological analysis
• resample methods producing daily, monthly and quarterly summaries
• Rolling windows for moving averages and short-term performance signals
• shift method calculating period-on-period changes and growth rates
• Cumulative measures for year-to-date totals and target tracking
• Missing-date detection within expected reporting calendars and sequences

Participants can identify unusual records and distinguish data errors from meaningful exceptions.

• Distribution summaries using quantiles, percentiles and interquartile ranges
• IQR rule for identifying potential numerical outliers
• Z-score interpretation for standardised deviation screening
• Segment-level thresholds preventing misleading global outlier comparisons
• Exception flags built from policy limits and analytical conditions
• Investigation queue prioritised by materiality, recurrence and operational risk

Participants can select and produce charts that communicate analytical findings accurately.

• Chart selection matched to comparison, composition, distribution and trend questions
• Pandas plot methods for rapid exploratory visualisation
• Matplotlib figures, axes, titles, labels and legends configuration
• Bar charts ordered clearly for category comparison and ranking
• Line charts showing time-series movement, targets and reference lines
• Chart review for scale distortion, clutter and misleading encodings

Participants can prepare readable analytical outputs for stakeholders and downstream systems.

• Formatted reporting tables using round, rename and column ordering
• Styler methods for number formats, highlights and conditional presentation
• ExcelWriter exports with multiple sheets and controlled output names
• CSV exports preserving encoding, delimiters and required field order
• Output folders organised by run date, version and reporting cycle
• Report notes stating definitions, source periods and material limitations

Participants can organise repeatable data processing into readable and maintainable code.

• Functions separating import, cleaning, transformation and reporting tasks
• Parameters controlling file paths, dates, thresholds and output locations
• Method chaining with pipe for readable transformation sequences
• Reusable validation functions returning exceptions and control metrics
• Configuration dictionaries storing mappings, rules and reporting settings
• Code comments explaining assumptions rather than restating syntax

Participants can make analytical scripts safer to rerun and easier for colleagues to review.

• try-except blocks handling missing files and conversion failures
• Assertions checking expected columns, row counts and value ranges
• Unit-test concepts for functions containing business-critical calculations
• Random seeds ensuring repeatable samples and generated test data
• Requirements files recording package versions for shared environments
• Run logs capturing inputs, outputs, warnings and processing timestamps

Participants can present findings with appropriate evidence, limitations and recommended actions.

• Analytical question framed with decision owner, measure and time period
• Finding statements separating observed evidence from interpretation and recommendation
• Control totals supporting confidence in reported analytical results
• Limitations disclosed for missing data, assumptions and coverage gaps
• Stakeholder review questions testing relevance, clarity and actionability
• One-page analysis brief combining question, method, findings and actions

Participants can build, review and present a complete Pandas analysis that converts raw operational data into validated findings and decision-ready outputs.

• Business case requirements translated into analytical questions and success measures
• Raw source files imported, profiled and documented in a notebook
• Cleaning rules applied to missing values, categories and identifiers
• Joined datasets reconciled through key checks and control totals
• Grouped measures, time trends and exceptions analysed against objectives
• Charts and reporting tables prepared for a defined stakeholder audience
• Peer review assesses reproducibility, assumptions, findings and recommended actions