Articles by Hafizh Yuwan Fauzan
- Claude in Excel and PowerPoint: From Portfolio Data to a Board Deck: A worked session: Claude in Excel builds the analysis and slide storyline, Claude in PowerPoint builds the board deck, then both handle a data correction.
- Claude in Excel: From a Messy Loan Export to a Risk Dashboard: A worked session with Claude in Excel: cleaning a messy loan export, building a risk dashboard, a round of manager feedback, and the silent error it traced.
- How I Use Claude in PowerPoint to Build Decks Faster: How I turn finished credit risk analysis into committee decks with Claude in PowerPoint: setup, three workflows, prompts to copy, and where it needs a human.
- consulting-pptx-skill for Claude Code: A Board Deck With a Built-In Reviewer: I rebuilt a Q3 board deck with consulting-pptx-skill for Claude Code: a slide rulebook, two checkers and a fresh-eye reviewer. What it caught and missed.
- Claude Code Superpowers: Building a PSI Monitor the Careful Way: The Superpowers plugin makes Claude Code ask, plan, test and debug before it codes. A worked session building a monthly PSI and CSI monitor on synthetic data.
- Setting a Scorecard Cutoff: Gains Tables, Break-Even Odds and Strategy: Part 6 of the credit scorecard series: reading a gains table, the approval vs bad rate trade-off, break-even cutoffs, risk-based score bands and overrides.
- Defining 'Bad' in Credit Scoring: Samples, Windows, Roll Rates and Segments: Part 2 of the credit scorecard series: who belongs in the sample, how long to watch each account, and how roll rates and vintages decide what counts as bad.
- How a Credit Scorecard Works: Points, Odds and Why Lenders Use Them: Part 1 of a practical series on credit risk scorecards: how points add up to odds, the five kinds of scorecard, and what to settle before any data is pulled.
- After Go-Live: Scorecard Monitoring, PSI, PD and Expected Loss: Population and characteristic stability, back-end monitoring, when to recalibrate, and turning a credit score into PD and expected loss. Part 7 of 7.
- Reject Inference: Why Approved-Only Scorecards Are Too Optimistic: Part 4 of the credit scorecard series: why approved-only data understates risk, the main reject inference techniques, parcelling worked through, and how bureau data helps.
- From Model to Scorecard: Scaling Points, KS, Gini and Validation: Part 5 of the credit scorecard series: turning a logistic regression into points, then judging the scorecard with a confusion matrix, KS, Gini, divergence and out-of-time tests.
- Weight of Evidence, Information Value and Logistic Regression for Credit Scorecards: Part 3 of the credit scorecard series: cleaning data, weight of evidence, fine and coarse classing, information value, and turning it all into a logistic regression.