Payment Risk and Payment Fraud: Data Science and Analytics

Learn payment basics, including open loop vs closed loop systems, push vs pull payments, and methods like cards, ACH, checks, cash, wire, and PayPal.

Learn the card system's buyers, issuer, card network, acquirer, and merchant, how authorization and clearing work, and the flow of fees and risks for card present and not present transactions.

Learn how chargeback and refund differ in the card system, who initiates each, and why merchants prefer refunds, including fraud programs, 3DS2.0 liability shifts, and in-house risk teams.

Explain the ACH system, its roles (buyer, ODFI, ACH operator, RDFI) and the debit and credit transactions, plus cost, same-day ACH, and risks of insufficient funds and fraud.

Learn about payment risk and fraud across buyers and sellers, including account takeover and non-sufficient funds, and how 3DS2.0 and PSD2 regulate this landscape.

Examine consumer risk in payments by covering account takeover, stolen financial, non sufficient funds, and family fraud, and learn how unauthorized access and credential theft enable ATO.

Learn how stolen financial information enables fraudsters to use card or bank data for online purchases, and examine non sufficient funds scenarios, data breaches, malware, and fraud signals.

Identify family fraud by analyzing signals from account profiles, purchase patterns, and linkages; expect long-tenure accounts using the same card on the same device, unusual purchases, and low cross-account linkages.

Explore merchant risk by distinguishing credit risk from fraud risk, and learn how bankruptcy, items not received (INR), and significant not as described (SNAD) affect payments.

Identify merchant risk by analyzing GMV trends, delivery days, and segments (young, SMB, large, too big to fail), use signals such as account history and user feedback to detect fraud.

Define hypothesis testing with null and alternative hypotheses, learn how to use rejection rules in fraud detection, and explore its role in modeling, confusion matrices, and AB testing.

Explore linear regression fundamentals, including simple and multivariable models, ordinary least squares, slope and intercept, residuals, and minimizing sum of squared errors to predict outcomes.

Explore how r-squared, adjusted r-squared, and p-values assess linear regression models, compare them to the mean model, and interpret f values for statistical significance.

Build a linear regression in Excel by plotting years of experience (X) against salary (Y), using the linear estimation function to obtain slope, intercept, and R squared, and predict salary.

Explore logistic regression fundamentals, where y is a probability, link predictors to probabilities via log transformations, and distinguish binary, multinomial, and ordinal forms from linear regression.

Learn how to estimate logistic regression parameters using maximum likelihood, compare models with McFadden's pseudo r-squared, and compute the r-squared and p value via the chi-squared test to assess significance.

Explore how a decision tree uses nodes to split data, with age over 60 and saving over $70,000, and discuss its basis in machine learning models and the outfitting issue.

Build and evaluate a fraud-detection decision tree using purchase velocity and account age, applying Gini impurity to split nodes and assessing performance with a confusion matrix.

Learn how gradient boosting builds a regression model by starting with an initial prediction, iteratively fitting trees to residuals, applying a learning rate, and combining trees for improved profit predictions.

Explore Xgboost, the extreme gradient boosting model, its relation to gradient boosting and decision trees, and its efficiency with super large datasets and computing power.

Builds an Xgboost regression tree by calculating residuals, similarity scores, and gains to choose optimal splits, applies regularization to prune branches, and updates predictions with the learning rate.

Discover how the SQL select distinct statement returns only unique values, including when used with aggregate functions, with a concrete example selecting distinct emails from a person table.

Learn to filter data in SQL with the where clause, using syntax and conditions such as salary greater than 100K and major equals statistics from the employee table.

Explore how the SQL join clause combines data from multiple tables using inner, left, right, and full outer joins, including self joins, with practical examples.

Explore how the in operator works in sql, where it serves as a shorthand for multiple or conditions or with subqueries, illustrated by a location table example.

Discover how the not equal operator in SQL filters data by excluding a value, shows two equivalent forms, and applies to a Location table example not equal to US.

Learn how the datediff function in SQL computes differences between dates, using day, week, month, or year intervals, to filter transactions in the last 30 days.

Learn how the case when statement in SQL evaluates conditions to return values, illustrated by a transaction example where approval occurs only when both model and manual decisions are approved.

Master how to convert data types in SQL with the cast function. Apply cast to change salary to number and enrollment timestamp to date, with practical syntax and examples.

Explore how window functions operate in SQL using the over clause, including partition by and order by, and compare row_number, rank, and dense_rank with a third-highest salary example.

Explore subqueries in SQL, using the where clause and in operator to restrict results across multiple tables, and compare subqueries with joins through practical examples.

Learn to perform a complex join across two tables using group by, window or aggregate functions, count votes, order by, and limit to identify winning candidate and reveal the name.

Learn how to apply join and aggregate functions to combine tables, group by department, and count people per department using the person and department tables.

Learn to use having and where together to filter movie sales by date, group by movie id, sum quantities, and apply having over ten copies sold.

Learn how to find and remove duplicates using window functions, partition by email, order by id, and row number, with alternatives like having, join, and group by.

Explore how to solve nth number problems in SQL using window functions, limit and offset, and max, with a clear ranking approach to find the second highest height.

Learn to compare data from previous dates and records using self joins and window functions, with weather and transaction examples illustrating partition by and order by concepts.

Boost SQL query efficiency by reducing dataset size, using targeted where filters, and selecting specific columns. Understand the SQL execution order to improve day-to-day query performance.

Explore core Python concepts: statements, indentation, and comments, with examples using variables like a=1, line continuation, semicolons, and code blocks in Jupyter Notebook.

Explore Python data types, including numbers and strings with integer, float, and complex subtypes, plus lists, sets, tuples, and dictionaries, mutability and order, and type conversion with int() and float().

Define and call Python functions to perform tasks and reuse code, using def function_name(parameters) and indentation; explore nested functions through a greet example that prints a personalized message.

Explore Python operators, including the remainder of division, the rounding and exponent operators, comparison and logical operators, and assignment plus in and not in membership tests.

Learn how to use Python if else to make logic judgments with if, elif, and else, including nested conditions, and apply these patterns to data scenarios in payment risk analytics.

Explore Python's while loop, its syntax and condition-driven execution, and compare it with the for loop. Learn when to use while loops and how break affects control flow.

Master indexing and slicing in Python lists, including negative indices; understand mutability via assignment, and remove items via delete, remove, or pop, including clearing the list.

Master Python sets by adding and removing elements with add, update, discard, remove, and pop; explore union, intersection, difference, and symmetric difference.

Explore Python dictionary basics, including unordered key-value pairs, curly bracket syntax, access by keys, mutability, no duplicates, and versioned ordering since Python 3.7.

Learn how to manipulate Python dictionaries: add and update keys, remove entries with delete or pop, and compare dictionaries using equal and not equal operators, demonstrated in a Jupyter Notebook.

Master numpy 101 and learn how arrays enable fast, memory-efficient data operations with homogeneous types, using numpy as np to create and access elements by index or slice.

Learn to reshape a 1D numpy array into a 2D array with rows and columns, index and iterate elements, sort rows, and filter with boolean indexing in a Jupyter Notebook.

Master Python Pandas 101 to explore DataFrame structures, indexing, and how to read data from CSV and JSON files for data analysis.

Learn to access data in a pandas data frame using head, locate, and column names to select height and weight in a Jupyter notebook.

Explore a dataframe using info, length, and columns, then filter with loc and sort with sort_values in pandas. Practice applies to height and weight data.

Master data pre-processing in a dataframe with Pandas 401, handling missing values, correcting data types, and removing duplicates using drop and fill techniques.

Learn how to join dataframes in pandas with the merge function, perform inner and left joins on shared columns, and review calories, weight, and height examples.

Explore matplotlib, a powerful Python library for data visualization that supports plots, histograms, 3D surfaces, and other charts with various settings; learn its easy syntax.

Explore matplotlib 201 basics: plot data with lines or dots, create histograms from random data with mean 100 and standard deviation 40, and display charts in Jupyter notebooks.

Apply a three-step data thinking process: reveal overall trends with sales and price, dive into segmentation by house type and repeat sales, then present data-driven recommendations using statistics and SQL.

Explore the Nashville housing trend by querying data with sql and aggregating gmv, counts, and average price, maximum price, and minimum price across house sale, house details, and seller information.

Query Nashville housing data with sql by joining house, house details, and seller tables to compute sales metrics. Identify areas, price versus value trends by bedroom count and build year.

Analyze two years of revenue and cost data using SQL to compute net revenue, identify drivers, and present actionable solutions, highlighting fraud cost reductions as the main revenue driver.

Build a Python decision tree to predict default and reduce fraud costs, using features like age and linking accounts, and quantify impact on net revenue.

Master the fundamentals of payment risk with data science and analytics, applying statistics, modeling, machine learning, sql, and python to analyze data and build models through two case studies.